MIT SCM Symposium 2024: Full Schedule

Welcome to the MIT SCM Research Symposium 2024 and thank you for joining us! We hope you find the day exciting and informative. To enhance your experience, we are utilizing this platform in order to help identify projects of top interest to you. Below you will find not only the schedule, but also the abstracts and attached executive summaries of each project. Additionally, you can better acquaint yourself with the student presenters and advisors through their profiles. Thank you again and we look forward to hosting you.

8:00am EDT

Breakfast

Friday May 17, 2024 8:00am - 9:00am EDT
Grand Ballroom Foyer 40 Edwin Land Blvd, Cambridge MA 02142, USA

8:45am EDT

Welcome Remarks

Friday May 17, 2024 8:45am - 9:00am EDT
Grand Ballroom 40 Edwin Land Blvd, Cambridge MA 02142, USA

9:00am EDT

Forecasting Drilling Bits Demand: A New Horizon for an Energy Global Technology Company’s S&OP

In response to the challenges of forecasting demand for drilling bits, this capstone project aimed to enhance the accuracy and efficiency of demand prediction for a global technology company in the energy sector. The goal was to replace outdated manual forecasting methods with automated causal and time-series models, optimizing the company’s Sales and Operations Planning (S&OP) processes in preparation for larger automation within the company’s Integrated Business Planning software (IBP). Utilizing historical data on rig counts and market shares, these models predicted future bit runs and estimated associated revenues with significantly improved accuracy, reducing global Mean Absolute Percentage Error (MAPE) from 6–8% to 2–3%. The analysis revealed that the performance of these models varied considerably among different geographic units (“geounits”), highlighting the necessity for customized forecasting strategies. Importantly, the best-performing models correlated with geounits’ business models, whether rental or bulk sales. Simpler time-series models frequently outperformed more complex causal models, suggesting that complexity does not always yield better forecasting results. This project streamlined the forecasting process and laid a strong foundation for ongoing improvement and adaptation to emerging market conditions. The integration of these models into the company’s S&OP practices represents a significant step forward in leveraging technological advancements to maintain a competitive edge in the energy sector.

Student Presenters

Thiago Faury

Materials & Supplies Coordinator, SLB

Advisors

Ilya Jackson

Postdoctoral Associate, MIT Center for Transportation and Logistics

Faury Forecasting Drilling Bits Demand Executive Summary pdf

Friday May 17, 2024 9:00am - 9:30am EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

DMP-Demand Planning

9:00am EDT

Optimizing Tempur Sealy International's North American Distribution Network for the Adjustable Base Category

Rising inventory costs and lower inventory turnover due to long lead times and supply/demand volatility have led to a need for our sponsor company, Tempur Sealy International, to reassess its sourcing, inventory, and deployment strategy for its adjustable base category. We develop a linear program optimization model capable of analyzing various network configurations, comparing direct-to-distribution center and hub & spoke models, and sourcing decisions, comparing offshore and nearshore suppliers, to define the optimal supply chain network. We conduct scenario analysis using a Monte Carlo simulation to understand how various design decisions affect cost and service. Additionally, we leverage our model to execute forward-looking supply planning for various scenarios, using demand forecasts for the next year to identify opportunities for nearer-term cost savings opportunities. Our capstone provides Tempur Sealy with actional insights for quantifying the cost and service impacts of inventory decisions, network flow paths, and supplier sourcing decisions.

Student Presenters

Sophie Hsu

MIT SCMr student with a background Tech supply chain, specifically in material planning and program management.

Drake Turnquist

MIT SCMr student with a background in digital supply chain consulting focused on large scale transformation and logistics technology projects.https://www.linkedin.com/in/draketurnquist/

Advisors

Matthias Winkenbach

Director of Research, Massachusetts Institute of Technology

Dr. Matthias Winkenbach is a Principal Research Scientist at MIT, where he serves as the Director of Research of the MIT Center for Transportation & Logistics. In this role, he manages CTL’s research activities and shapes the center’s strategic agenda of creating supply chain... Read More →

Selene Silvestri

Research Scientist, Center for Transportation and Logistics, Massachusetts Institute of Technology

I am a Research Scientist at the MIT Center for Transportation & Logistics. My current research spans across the area of supply chain network design and optimization. My work is performed in collaboration with global organizations, and it aims to help such organizations improve decision-making... Read More →

Executive Summary Optimizing Tempur Sealy International’s North American Final pdf

Friday May 17, 2024 9:00am - 9:30am EDT
Skyline AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

NTD-Network Design

9:00am EDT

Strategy Formulation for SKU Rationalization using Financial and Bill of Material Metrics

Newell Brands, a conglomerate with a diverse product portfolio, has traditionally relied on financial metrics for SKU rationalization. Now, they are seeking a SKU rationalization strategy for “Pen” portfolio to eliminate SKUs that add more complexity to the Bill of Material (BOM) with least value creation. How can Newell Brands use a combination of financial and BOM metrics to infer the criterion for SKU rationalization? The methodology involved data analysis, relevant metric selection, development of a rating system, and a three-step procedure for SKU rationalization. We obtained results that adjusting the ratings for each metric can help optimize the product portfolio in accordance with business requirements. The methodology can be extended to other product portfolios too. Conclusions highlight the effectiveness of the methodology in identifying SKUs for rationalization and the importance of considering unique components in the analysis.

Student Presenters

Seyeon Park

Graduate Student, Massachusetts Institute of Technology

Mayank Raj

Student, MIT

Advisors

Milena Janjevic

Research Scientist, Megacity Logistics Lab, MIT Center for Transportation & Logistics

Friday May 17, 2024 9:00am - 9:30am EDT
Longfellow AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SCS-Supply Chain Strategy

9:00am EDT

Truckload Procurement: From State-of-the-Practice to State-of-the-Art

Truckload procurement practices vary widely across industry and firm spend levels. This research utilizes a framework to identify shipper behaviors that represent state-of-the-practice in truckload procurement while also highlighting shipper behaviors that represent state-of-the-art. The presented anlaysis uses data collected from 1) a survey of 300 shippers and 2) semi-structured interviews with nearly 50 shippers. Is your firm with the herd or on the cutting edge of practices in truckload procurement?

Student Presenters

Maria Lucchi

Student

Advisors

Chris Caplice

Executive Director, MIT CTL

Dr. Caplice serves as the Executive Director of the Massachusetts Institute of Technology’s Center for Transportation & Logistics (CTL) where he is responsible for the planning and management of the research, education, and corporate outreach programs for the center. He created... Read More →

Friday May 17, 2024 9:00am - 9:30am EDT
Longfellow C 40 Edwin Land Blvd, Cambridge MA 02142, USA

TRS-Transportation

9:30am EDT

Machine Learning in Prescribing Optimal Target Inventory Levels for FMCG

For our project sponsor, a leading global Fast Moving Consumer Goods (FMCG) company, the goal goes beyond simply meeting complex demands. The focus is on optimizing inventory management strategies to balance the crucial trade-off between avoiding stockouts and minimizing the costs associated with excess inventory. Our project explores machine learning's potential to enhance the company's strategy for setting target inventory levels. We aim to use machine learning models to determine how much optimization is possible, ensuring that product service levels are sufficient to meet demand. Utilizing company data and external variables as features, we developed and evaluated various machine learning techniques by tuning different hyperparameter sets to identify the most accurate binary classification model. This model is designed to accurately predict stockout events for each stock keeping unit (SKU) across the company's warehouses. We selected the most accurate model based on a classification report that included key metrics such as precision, recall, f1-score, and accuracy. Using this model, we conducted sensitivity analysis to test different scenarios of adjusting current target inventory levels and assess how these changes could affect predicted stockout events. The findings will offer the company recommendations on setting inventory levels for specific SKUs at warehouse sites, ensuring that target service levels are maintained.

Student Presenters

Firanza Fadilla

Wenchao Zhang

Advisors

Devadrita Nair

Postdoctoral Associate, MIT CTL

Ilya Jackson

Postdoctoral Associate, MIT Center for Transportation and Logistics

Friday May 17, 2024 9:30am - 10:00am EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

MAL-Machine Learning

9:30am EDT

Criticality of U.S. Food Supply Chains from Latin America

Food imports from Latin America via ocean transport are important for the U.S. food supply and economic growth. The resilience of the U.S. port network is fundamental in maintaining this supply. A disruption in one or more ports could compromise the network’s capacity to keep a smooth flow of perishable goods before their expiration date. This study focuses on seven critical perishable goods, including dry, cold, and frozen containerized cargo. It first develops a network analysis for each node and path, identifying the centrality and criticality of primary ports and maritime routes within the U.S. Furthermore, it explores the potential consequences of disruption in these primary ports through simulation and proposes contingency strategies. The results rank the most critical ports by node degree, betweenness and closeness centralities, as well as the volume, and value of goods received. Additionally, the simulation displays the reallocation of bananas to other viable ports in the event of partial or total reductions of operations at a critical port. Such an analysis offers alternative solutions within the U.S port network for vessels whose original destinations experience disruptions.

Student Presenters

Santiago Perez

Center For Transportation and Logistics, MIT, U.S.A.

Mateo Rojas

Student, Center For Transportation and Logistics, MIT, U.S.A.

Advisors

Chris Mejia

Director, MIT SCALE Network, MIT Food and Retail Operations Lab

Christopher Mejía Argueta is a Research Scientist at the MIT Center for Transportation and Logistics. He develops applied research on retailing operations and food supply chains for multiple stakeholders including consumer packaged goods manufacturers, carriers and retailers in the... Read More →

Elenna Dugundji

Research Scientist, MIT Center for Transportation and Logistics

Friday May 17, 2024 9:30am - 10:00am EDT
Longfellow AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SCS-Supply Chain Strategy

9:30am EDT

Dell - Supply Chain Emissions Hotspot Analysis

In this capstone project, the complex relationship between Dell Technologies and its suppliers is examined, with a focus on how the suppliers’ supply chain impacts Dell's greenhouse gas (GHG) emissions. Facing the challenges of climate change, Dell is committed to achieving net-zero GHG emissions by 2050, with a significant focus on reducing emissions across its supply chain. The project aims to improve the accuracy of emissions data reported by suppliers and to identify inaccuracies, enabling Dell to effectively meet its sustainability targets. By incorporating an outlier detection model, deriving an academic model, and performing sensitivity analysis, the project assesses the impact of supplier decisions on Dell's Scope 3 emissions, contributing to strategic and collaborative efforts required to mitigate the environmental impact of the supply chain. This effort not only aids Dell in advancing toward its sustainability targets but also provides a model for other companies seeking to enhance sustainability within their supply chains.

Student Presenters

Hansika Gupta

Dhorn Kosolpatanadurong

Advisors

Josue Velazquez Martinez

Research Scientist, MIT Sustainable Supply Chain Lab

Josué C. Velázquez Martínez is a Research Scientist, and Lecturer at the MIT Center for Transportation and Logistics, with focus on Logistics and Supply Chain Management in transportation, manufacturing, and retail industries, and has more than 10 years of experience in conducting... Read More →

Friday May 17, 2024 9:30am - 10:00am EDT
Skyline AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SUS-Sustainability

9:30am EDT

Accounting for Uncertainty: An Empirical Analysis of Truckload Budgeting

This study investigates the United States full truckload procurement process and persistent issue of budget overruns faced by shippers during their budgeting. Despite planning through Requests for Proposals (RFPs) that forecast shipping volumes and secure contractual rates with carriers, shippers regularly confront unplanned expenses surpassing their budgets. The key problems addressed are the discrepancy between planned budgets and actual expenditures, accentuated by the dynamics of the spot market and the unpredictability of freight volumes. Utilizing data from 13 shippers’ Transportation Management Systems (TMS), provided by C.H. Robinson's TMC division, this research delves into the factors contributing to budget overruns. The analysis covers 196 transportation procurement events across 13 shipper companies over six years, highlighting a consistent trend of budget overruns, with some instances reaching up to 280%. The methodology employed includes a quantitative model that helps shippers plan their budgets. The findings underscore the necessity for a more sophisticated budgetary framework that integrates spot market factors, enabling shippers to more accurately anticipate and manage transportation costs. This study not only sheds light on the complexities of freight budgeting but also suggests a shift towards incorporating data analytics to enhance forecasting accuracy and budget reliability.

Student Presenters

Shayna Moliver

Mostafa Taheri

Student

Advisors

Chris Caplice

Executive Director, MIT CTL

Friday May 17, 2024 9:30am - 10:00am EDT
Longfellow C 40 Edwin Land Blvd, Cambridge MA 02142, USA

TRS-Transportation

10:00am EDT

Buying Channels strategy: Advanced data analytics for procurement efficiency

There is a growing need to leverage data for improvement of processes in the procurement organization. Understanding the interaction of stakeholders with procurement structures and processes through advanced data analytics benefits the entire organization as significant efficiency and optimization can be realized.
This capstone project explores the procurement buying channels process in the sponsor company with the aim of providing insights on variables that influence the use of those buying channels by internal customers and improving the channel planning process. Clustering for the features was carried out using K-means algorithm. Regression and Classification Machine learning models were then built to identify the most relevant variables in the use of buying channels, and to predict buying channel behaviors.
Implementing this approach to data analysis revealed that predicting the completion time of purchases through various channels had low accuracy due to significant variability in cycle times Additionally, there was confusion in channel usage for similar purchase categories, leading to misclassification of channels. This underscores the necessity for a clearer remapping of channels and the empowerment of requesting parties to better understand and utilize the process.

Student Presenters

Foyinsola Adeyemi

Fabrizio Boaron

I have worked in investment banking until 2014, when, together with two friends, I founded and run an investment advisory company. In 2017 I moved to my current role as portfolio manager for a private equity firm based in London. I am now responsible for structuring acquisitions of... Read More →

Advisors

Maria Jesus Saenz

Executive Director, MIT SCM Master and Director, MIT Digital SC Lab, MIT Center for Transportation and Logistics

Dr. Maria Jesus Saenz is the Director of the research area on Digital Supply Chain Transformation at the MIT Center for Transportation and Logistics, as Research Scientist. The primary research aims at leveraging the connections among inter-organizational business drivers when facing... Read More →

Friday May 17, 2024 10:00am - 10:30am EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

DAT-Data Analytics

10:00am EDT

Fresher Food Supply: Evaluating the impact of pick-to-zero strategy on freshness of produce using discrete event simulations

Consumers frequently reach behind supermarket shelves to find products with the latest expiration dates. This instinctive behavior highlights a universal desire for fresh produce. Therefore, our capstone sponsor is eager to consistently deliver fresh produce to maximize customer satisfaction. One innovative approach to increasing freshness is to minimize the duration produce spends in the supply chain by implementing a pick to zero strategy. However, can a pick to zero strategy improve freshness without significantly increasing costs or waste? Using discrete event simulation in SimPy we compared the current state and future state supply chain by incorporating factors of uncertainty at each node of the network. The simulation model indicates that a pick to zero strategy improves freshness of produce by 16%, reduces excess by 40% but doubles the transportation costs. However, using a strategically located consolidation center in conjunction with the pick to zero strategy reduces the transportation costs. Furthermore, our results also indicate that products with higher forecast accuracy and lower forecast variance are more suitable for a pick to zero strategy.

Student Presenters

Shantanu Baviskar

Kalyan Simha

Advisors

Chris Mejia

Director, MIT SCALE Network, MIT Food and Retail Operations Lab

Eva Ponce

Director, Omnichannel Distribution Strategies, MIT Center for Transportation & Logistics

Dr. Eva Ponce is the Director of the research area on Omnichannel Distribution Strategies at the MIT Center for Transportation & Logistics, as Research Scientist. Her current research focus is the design of omnichannel distribution strategies that integrates online and offline channels... Read More →

Friday May 17, 2024 10:00am - 10:30am EDT
Longfellow AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

ROP-Retail Operations

10:00am EDT

Optimizing the Life Cycle of Last-Mile Packaging

More than 85 million tons of cardboard waste are created annually, with most ending up in landfills. This capstone project aims to develop a mathematical optimization model to help last-mile delivery companies reduce their carbon dioxide (CO2) footprint by collecting and reusing cardboard cartons. Specifically, the optimization model suggests from which customers cardboard boxes should be collected during a delivery route, considering practical constraints such as limited vehicles' capacity and maximum driving time. Our results indicate that collecting all potential cardboard boxes is not imperative to achieve the highest possible reduction in CO2 emissions.

Student Presenters

Cristina Aguirre

Omar Pineda

Advisors

Matthias Winkenbach

Director of Research, Massachusetts Institute of Technology

Selene Silvestri

Research Scientist, Center for Transportation and Logistics, Massachusetts Institute of Technology

Friday May 17, 2024 10:00am - 10:30am EDT
Skyline AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SUS-Sustainability

10:00am EDT

Predicting the Likelihood of a Shipment Write-Off

In the U.S. trucking industry, freight brokerages act as vital intermediaries between shippers and carriers, but they face financial risks due to write-offs from unpaid services. Despite the recognized importance of mitigating these financial risks, the sponsoring company does not currently have a predictive model to assess the likelihood and magnitude of write-offs, making it challenging to prevent financial losses before they occur. This study tackles this issue by analyzing shipment data and historical write-off incidents to identify key predictors of financial write-offs. Utilizing logistic and linear regression models, it quantifies the risk associated with each shipment, enabling the brokerage to prioritize transactions with lower risk profiles. The analysis revealed that specific shipment characteristics, such as mode of transportation, significantly influence the likelihood and magnitude of write-offs. Predictive models developed in this study were able to accurately forecast the probability of write-offs, offering a tool for more informed decision-making. The findings demonstrate the potential for predictive modeling to significantly reduce financial risks for freight brokerages by enabling preemptive identification of high-risk shipments. By applying this predictive approach, freight brokerages can enhance their financial stability and operational efficiency, contributing to the overall health of the trucking industry's economic ecosystem.

Student Presenters

Oscar Bonet

Joshua Weston

Advisors

Chris Caplice

Executive Director, MIT CTL

Devadrita Nair

Postdoctoral Associate, MIT CTL

Friday May 17, 2024 10:00am - 10:30am EDT
Longfellow C 40 Edwin Land Blvd, Cambridge MA 02142, USA

TRS-Transportation

10:30am EDT

An Application of Neural Networks to improve Forecast Accuracy in a FMCG company

FMCG companies are facing the dilemma of improving customer service while keeping out-of-stock and reducing supply chain costs. One way to overcome this is to improve demand forecast accuracy. Whereas achieving this improvement is critical for business success, and using state-of-the-art methods should be highly spread among companies, most FMCG companies still rely on more traditional approaches. AB InBev is trying to upgrade its forecasting methods. The company has reached a maturity level in demand forecasting and uses a mix of time series and machine-learning models. Nevertheless, this hybrid approach has obtained only 60% of demand forecast accuracy in Colombia's market leaving room for improvement. Moreover, AB InBev bought “BEES,” an e-commerce platform for its business-to-business channel, and expects to use BEES’ data as the main input in its demand forecasting process. In this capstone project, we made a proof of concept for two aspects of interest for AB InBev: (i) Is BEES data mature enough to be useful in a demand forecasting context, and (ii) what is the expected forecast accuracy lift obtained using neural networks as a demand forecasting technique. For the latter, we applied a global Sequence-to-sequence recurrent neural network model to forecast the demand of 289 different SKUs-DC combinations. Based on our analysis, we provide several recommendations to improve demand forecasting accuracy successfully.

Student Presenters

Diego Campos Durand

MIT SCM 2024, MIT CTL

Miguel Iturralde

Advisors

Jafar Namdar

Postdoctoral Associate, MIT Center for Transportation and Logistics

Friday May 17, 2024 10:30am - 11:00am EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

DMP-Demand Planning

10:30am EDT

Dealing with Supply Chain Complexities with Scenario Intelligence

The recent escalation in the VUCA (Volatility, Uncertainty, Complexity, and Ambiguity) world is adding unprecedented levels of complexity to businesses. Philip Morris International is no exception. Factors such as geopolitical conflicts, raw material shortages, rising inflation, and changes in country legislation are exerting pressure on PMI's supply chain. This pressure, combined with PMI's portfolio expansion driven by a strategic shift in recent years has created the necessity for managing risk uncertainties in the supply chain through Scenario Intelligence. This project will employ Simulation Modeling and Scenario Planning as our primary methodologies to address the challenges with complexities.

Student Presenters

Zhan Ding

Luis Felipe Sena

Advisors

Devadrita Nair

Postdoctoral Associate, MIT CTL

Maria Jesus Saenz

Executive Director, MIT SCM Master and Director, MIT Digital SC Lab, MIT Center for Transportation and Logistics

Friday May 17, 2024 10:30am - 11:00am EDT
Longfellow AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SCS-Supply Chain Strategy

10:30am EDT

Achieving Operational Excellence by Ensuring Optimization of Electric Vehicles vs. Internal Combustion Engines in Fleet Vehicles

Recognition of the need to reduce greenhouse gases and carbon footprints has led us to investigate what actions are available to companies that rely on fleets for business purposes. There is a strong alignment for the collective contributions of all actors, i.e. nations, firms, and individuals, to limit the annual warming to below 2 degrees Celsius. Our sponsor company, with a global fleet of over 25,000 vehicles primarily comprised of Internal Combustion Engine (ICE) vehicles, is committed to significantly decarbonizing its fleet by 2030 to mitigate its CO2 emissions footprint and contribute to global warming reduction. This goal is to be achieved while maintaining operational excellence and within the company’s economic and operational constraints. To this end, our study first identified optimal locations for transitioning fleets from ICEs to Electric Vehicles (EVs), considering the geographical scope of the 50 US states plus the District of Columbia. Using Machine Learning Clustering techniques, we included endogenous factors (age of fleet, number of vehicles ) and exogenous factors (laws and incentives, temperature, gas price, and electricity price) to identify how to rank states according to their impact. Then, a logistic growth function, with a growth rate factor derived from 5 metrics, was applied to model the timing and strategy of EV implementation: Total Cost of Ownership (TCO), driving range, refueling, CO2 emissions, value-perception. We found that the adoption of EVs in a global corporation with a significantly large fleet is equally dependent on both endogenous and exogenous factors. Furthermore, to reap optimal benefits, the number of EVs in the company’s fleet mix should be gradually increased over the target period. Combining these 2 approaches allows the company to maintain control over operational performance objectives and predict future TCO and decarbonization implications. The model's applicability extends beyond the studied region to other geographical, political, and economic contexts, such as Europe or East Asia.

Student Presenters

Jospin Kalumire Mubaki

Niraja Shukla

Advisors

Elenna Dugundji

Research Scientist, MIT Center for Transportation and Logistics

Friday May 17, 2024 10:30am - 11:00am EDT
Skyline AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SUS-Sustainability

10:30am EDT

Cost to Serve

This project introduces a refined approach to cost allocation developed for our sponsor company, a manufacturer and commercializer of chemical solutions for the construction industry. Faced with the complexities of serving thousands of clients across North America, the company seeks to improve its volume-based allocation method to accurately identify profitable clients and enhance business decision-making. Utilizing the Shapley method from cooperative game theory, this study proposes a new model that incorporates geographic distances, shipment volumes, and other logistical factors into the cost allocation process. We employ Shapley values to assign costs based on the marginal contributions of each client to the overall transportation costs, a significant advancement over the volume-based proportional method. Initial results demonstrate the model's effectiveness in providing a fair and transparent cost distribution, supported by several methods of analysis which quantifies the improvement over the existing allocation policy. Furthermore, the implementation of this model was augmented through machine learning techniques, which enables predictive insights for cost allocation. This enhancement promises improved operational efficiency and strategic planning capabilities. The project not only addresses the immediate needs of our sponsor company but also sets a precedent for “cost-to-serve” applications in logistics-intensive industries.

Student Presenters

Matias Opazo

MIT SCM 24'

Prior to MIT, I led a 4,000+ container global logistics operation for a commodity trader. Operating with over 20 LATAM suppliers and customers worldwide. I hold a degree in Industrial Engineering from the Pontifical Catholic University of Chile.

Douglas Tjokrosetio

Management Consultant, Boston Consulting Group

Advisors

Milena Janjevic

Research Scientist, Megacity Logistics Lab, MIT Center for Transportation & Logistics

Friday May 17, 2024 10:30am - 11:00am EDT
Longfellow C 40 Edwin Land Blvd, Cambridge MA 02142, USA

TRS-Transportation

11:00am EDT

Decoding Carrier Preferences in Digital Freight: A Predictive Machine Learning Analysis

The freight brokerage industry is at a pivotal juncture, with digital platforms reshaping market dynamics and carrier preferences. This capstone project, undertaken in partnership with Nolan Transportation Group (NTG), employs a predictive machine learning model to decode and understand these evolving preferences. The study leverages a dataset comprising nearly 2 million brokerage transactions, enriched with comprehensive feature engineering, to model the likelihood of digital vs. traditional booking methods. The research uses advanced machine learning algorithms, especially Gradient Boosting with XGBoost, to identify key shipment characteristics that influence carriers' digital booking decisions; piercing through the complex interplay of shipment characteristics that decisively influence digital booking decisions. Central to these findings is the pivotal role of the time a load remains available on digital platforms in determining its likelihood of being digitally booked. The analysis underscores a critical insight: the probability of a load being booked digitally diminishes significantly with time, highlighting a narrow window for digital engagement. This discovery has valuable operational implications, suggesting a strategic shift towards minimizing internal competition for loads in the period of initial listing, thereby enhancing the effectiveness of digital channels. By offering a nuanced understanding of the temporal dynamics at play in digital freight booking, this research provides actionable strategies for fostering digital adoption and optimizing brokerage operations in the digital age. Through this lens, the study not only contributes to academic discourse but also equips industry practitioners with the insights needed to navigate the evolving landscape of freight brokerage.

Student Presenters

Perry Falk

EVP, Sales & Operations, Nolan Transportation Group (NTG)

Advisors

David Correll

Research Scientist, MIT Center for Transportation & Logistics

Dr. David Correll is a Research Scientist at the MIT Center for Transportation and Logistics, where he serves as a Course Lead in the MITx MicroMasters in Supply Chain Management program, and contributes to transportation research at Freightlab and MIT Sustainable Supply Chains. His... Read More →

Friday May 17, 2024 11:00am - 11:30am EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

MAL-Machine Learning

11:00am EDT

Impact of Exogenous Variables on AI/ML Forecasting Algorithm Prophet, in the FMCG Industry

As Artificial Intelligence (AI) and Machine Learning (ML) continue to advance, their application in supply chain demand forecasting has surged, significantly impacting the field. These ML techniques improve forecasting accuracy, ultimately supporting managerial inventory and planning decisions. In this project, we evaluate the underlying hypothesis that suggests the application of targeted exogenous variables as model features, would further improve the predictive accuracy of the popular ML algorithm “Prophet”. Provided with statistically normalized Fast Moving Consumer Goods (FMCG) sales data from the Indian Subcontinent, we investigated 5,184 data points from 2011 to 2023 across 36 products. We screened and shortlisted exogenous variables within economic, health, climate, and political areas and selected 10 for evaluation in the predictive model. Each product was cross-validated with variables built as univariate regressors (features), and it was found that the introduction of exogenous variables reduced the Mean Average Percent Error (MAPE) by as much as -40.93%. Furthermore, incorporating targeted exogenous variables together with Hyper-Parameter tuning (calibration), had astounding results, with MAPE decreases by as much as -143.48%. This study proves targeted exogenous variables beneficial for practical application in improving forecasting accuracy within the FMCG industry.

Student Presenters

Ryan Rocke

Head of Sourcing & Procurement, Western Hemisphere Chemicals, SLB

With over a decade of industry experience, Ryan brings a wealth of expertise in managing cross-continental supply chain teams across all of its pillars: sourcing, planning, procurement, manufacturing, warehousing, supplier management, logistics, reconciliation and field operations... Read More →

Lili Zhang

Student, Center For Transportation and Logistics, MIT, U.S.A.

I’m from China and have been working in the Oil and gas industry for the past 11 years. I have diverse backgrounds in O&G Engineering, manufacturing, and the Circular Economy. I’ve ventured into multiple supply chain domains, including procurement and sourcing, supply chain process... Read More →

Advisors

Ilya Jackson

Postdoctoral Associate, MIT Center for Transportation and Logistics

Maria Jesus Saenz

Executive Director, MIT SCM Master and Director, MIT Digital SC Lab, MIT Center for Transportation and Logistics

Executive Summary 2024 Lili Ryan pdf

Friday May 17, 2024 11:00am - 11:30am EDT
Longfellow AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SCS-Supply Chain Strategy

11:00am EDT

Innovative Green Network Design for a Multinational CPG Company

In response to global climate warming, corporations have solidified their sustainability commitments and intensified their efforts to reduce greenhouse gas (GHG) emissions. In partnership with a global CPG company, this work focuses on the feasibility of engaging wholesalers in the distribution network as third-party logistics providers (3PLs) to reduce emissions. Using shipment data provided by the company for a single area of the Kantō region in Japan, we calculated baseline emissions and costs across their current supply chain, in which wholesalers exist as customers. To assess the current network against the proposed network, in which wholesalers function as 3PLs, we built several proof-of-concept models using mixed-integer linear programming (MILP). Within the optimization models, emissions were calculated following both the Global Logistics Emission Council (GELC) Framework and the Network for Transportation Measures (NTM) methodology. We also explored the financial implications of the proposed network design by building MILPs to optimize cost. We expect these models to demonstrate that distribution through wholesalers will be the most optimal solution from an emissions perspective and provide insight into operationalizing this change. Our analysis could serve as a theoretical framework for utilizing the wholesaler network in operational scenarios, or as an input for future large-scale network optimization.

Student Presenters

Kavitha Balakumar

Fan Gruener

Student

Advisors

Josue Velazquez Martinez

Research Scientist, MIT Sustainable Supply Chain Lab

Sreedevi Rajagopalan

Postdoctoral Associate, MIT Center for Transportation and Logistics

Friday May 17, 2024 11:00am - 11:30am EDT
Skyline AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SUS-Sustainability

11:00am EDT

Truck Tetris: Optimizing Packaging for Transportation Efficiency

Consumer demand for frozen products from The J.M. Smucker Company has grown dramatically over recent years, driving an evident need to increase total ship output. The overarching problem that we attempted to solve in this project was how to maximize the amount of a specific frozen product loaded onto trailers in order to increase total ship output by changing carton, case, and pallet packaging dimensions and orientations without increasing the total number of trailers used. Empirical research was performed and through TOPS optimization modeling techniques, we modified carton, case, and pallet dimensions for 4-count, 10-count, and 18-count products. Informational interviews and a facility visit were performed to identify and further explore stakeholder objectives and success criteria. Upon conclusion, we provided a series of recommendations to maximize Smucker’s shipping output per trailer for the given product to reduce their transportation costs while meeting all of the requirements outlined by the identified stakeholders.

Student Presenters

Julia Mionis

Residential Student, Massachusetts Institute of Technology

AJ Shaw

Residential Student, Massachusetts Institute of Technology

Advisors

Miguel Rodriguez Garcia

Postdoctoral Associate, MIT Center for Transportation and Logistics

SCM13 Mionis Shaw ExecutiveSummary pdf

Friday May 17, 2024 11:00am - 11:30am EDT
Longfellow C 40 Edwin Land Blvd, Cambridge MA 02142, USA

TRS-Transportation

11:30am EDT

Lunch

Friday May 17, 2024 11:30am - 1:00pm EDT
Grand Ballroom 40 Edwin Land Blvd, Cambridge MA 02142, USA

1:00pm EDT

Impact of Downstream Supply Chain Dynamics on Patient Access to Oncology Medicines

Access to cancer medicines remains a significant challenge in many Low- and Middle-Income Countries (LMICs), limiting patients' ability to receive timely and affordable treatment. This study aims to analyze the impact of the pharmaceutical downstream supply chain on patient access to oncology medicines in LMICs. Utilizing a system dynamics approach, we developed a comprehensive causal loop diagram, CLD, to map the complex interactions between key variables and stakeholders in the downstream value chain. Qualitative data from interviews with PharmaCo experts and insights from a World Health Organization technical report were used to construct and validate the CLD. The analysis highlighted several key feedback loops influencing affordability and availability of cancer medicines, including market scale, competition, insurance support, and inventory management. Reinforcing and balancing effects among these loops were identified, highlighting the trade-offs and challenges in ensuring patient access. The study also examined the roles and interests of various stakeholders, such as manufacturers, distributors, healthcare providers, insurers, and governments. This research emphasizes the importance of a holistic, system-level understanding of the downstream supply chain dynamics to develop effective strategies for increasing patient access to life-saving cancer treatments in LMICs.

Student Presenters

Patricia Andrade

Ping Wang

Advisors

Jarrod Goentzel

Director, MIT Humanitarian Supply Chain Lab, MIT CTL

Friday May 17, 2024 1:00pm - 1:30pm EDT
Skyline AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

HEC-Healthcare

1:00pm EDT

Enhancing Efficiency in Cell & Gene Therapy Shipments: A Pathway to Scalability and Reliability

The logistics supporting life-saving Cell & Gene Therapy treatments, such as autologous CAR-T, face significant challenges due to strict constraints like time sensitivity, temperature control, and regulatory compliance. These constraints make the supply chain vulnerable to disruptions that could result in the loss or damage of these delicate, high-value therapies during global shipments handled by white-glove third-party couriers. To address this issue, a study was conducted to analyze historical shipment data from a qualitative and quantitative perspective. The goal was to create a model that could predict possible disruptions by calculating "validation points" throughout the shipment process and updating them continuously for each new input. As a result, the sponsors' planning team is informed in advance of any potential disruptions, allowing them to proactively reach out to their couriers for immediate action. Although still in its early stages, the model is providing more visibility into current processes and laying the foundation for a scalable solution to be implemented in the sponsor operating system. Additionally, this study has allowed the sponsor to review their current process with their couriers and identify areas for improvement in terms of process, data gathering, and data quality. A proposal for a roadmap is also included, outlining possible enhancements that could leverage Machine Learning and AI techniques.

Student Presenters

Kevin-Alexandre Jacquot

Advisors

Elenna Dugundji

Research Scientist, MIT Center for Transportation and Logistics

Thomas Koch

Postdoctoral Associate, MIT Center for Transportation and Logistics

Friday May 17, 2024 1:00pm - 1:30pm EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

RMG-Risk Management

1:00pm EDT

Capacity vs. Inventory: A Trade-Off Strategy for an FMCG Company

In the dynamic landscape of the fast moving consumer goods industry, companies must effectively manage production capacity and inventory to respond to fluctuations in demand. To mitigate the risk of lost sales, companies can adjust two levers to handle such volatility: increase inventory levels or production capacity. In this context, the planning team of our sponsor company, a large company in the snack food industry, aims to address this problem cost-efficiently, ensuring profitability amid uncertainty. They seek a tactical capacity-planning model that strikes the correct balance between inventory levels and capacity utilization. This paper takes a holistic approach to capacity and inventory management. We present a mixed integer linear programming model that handles the trade-off and balances holding inventory and increasing capacity on the tactical horizon. We model that trade-off using data from one of the sponsor company's production lines, the relevant stock-keeping units, and the associated inventory and capacity increase costs. While increasing capacity may seem counterintuitive because it entails lower utilization, our model shows that even a slight increase above the sponsor company's current baseline capacity can lead to lower inventory levels in specific periods, resulting in improved financial performance. Based on data from one year, our model recommends a 2.70% increase in production capacity, resulting in a reduction of 5.99% in inventory holding expenses and achieving an overall annual financial improvement of 0.53%.

Student Presenters

Jose Arturo Espinosa

Mohamed Gaafar

Advisors

Sean Willems

Capacity vs Inventory A Trade Off Strategy for an FMCG Company (Espinosa Gaafar) Executive Summary pdf

Friday May 17, 2024 1:00pm - 1:30pm EDT
Longfellow AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SCS-Supply Chain Strategy

1:00pm EDT

The Road Ahead: Leveraging Truckload Trends for Prescriptive LTL Heuristics

Anticipating fluctuations in Less-than-Truckload (LTL) volume presents challenges for shippers, carriers, and freight brokers alike. This capstone addresses this issue by developing a predictive model for LTL volume, leveraging insights gained from the cyclical nature of demand shifts between Truckload and LTL freight. Through analyzing various truckload metrics, this study identifies key indicators that precede changes in LTL volume. The resulting prediction model offers valuable insights and informs the creation of practical heuristics for freight brokers, enabling them to respond proactively to market fluctuations. By securing contract rates timelier, freight brokers can effectively navigate changing market dynamics, thus enhancing operational flexibility and adaptability within the industry.

Student Presenters

Bobby Kheny

Student, MIT

Claire Urbi

Advisors

David Correll

Research Scientist, MIT Center for Transportation & Logistics

Kheny & Urbi LTLModel ExecutiveSummary pdf

Friday May 17, 2024 1:00pm - 1:30pm EDT
Longfellow C 40 Edwin Land Blvd, Cambridge MA 02142, USA

TRS-Transportation

1:30pm EDT

Integrating Generative AI to Drive Efficiency in Spend Intelligence and Negotiation Strategy

This paper explores the integration of generative artificial intelligence (AI) technology into the procurement operations of a global healthcare company. Driven by a large procurement spend of $35BN with diverse sourcing information and massive amounts of data, the research aims to help our sponsor company develop a real-world proof-of-concept of generative AI that can be successfully implemented in the procurement function. With this objective in mind, we developed a chatbot that democratizes data mining skills to category managers and promotes smarter supplier negotiations. We implemented a retrieval augmented generation (RAG) approach which is less computationally expensive and reduces hallucinations. We used LangChain's text-2-SQL agent on the sponsor’s company relational database architecture. In parallel, we used LangChain's kuzuQAchain agent on a graph knowledge database architecture that we created using a Python library called Kuzu. The model takes the natural language queries and generates either SQL code or Cipher code (depending on the question type), retrieves the relevant information, and returns to the user's natural language with the answer to the prompt. We managed to develop a model that does not return false answers or hallucinations. The final prototype has been presented to the sponsor company and potential users who highlighted the promising benefits that this solution will offer when deployed in terms of efficiency and insight-gathering capabilities.

Student Presenters

Andres Ayala

Ria Verma

Advisors

Elenna Dugundji

Research Scientist, MIT Center for Transportation and Logistics

Thomas Koch

Postdoctoral Associate, MIT Center for Transportation and Logistics

Friday May 17, 2024 1:30pm - 2:00pm EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

DAT-Data Analytics

1:30pm EDT

Early Assessment of Economic Viability in Cell and Gene Therapy via Alignment of Cost and Market

Cell and Gene Therapy (CGT) treatment require significant upfront R&D investments coupled with lower patient demand volumes compared to traditional blockbuster drugs while not benefiting from the economies of scale principles for cost reduction. Therefore, ensuring the economic viability of a CGT treatment is one of the most pivotal considerations for pharmaceutical companies to balance between bringing life-saving innovation to the market and ensuring a sustainable business. In this work, we developed an integrated framework for early assessments of the economic viability of CGTs consisting of a high-level, manufacturer neutral cost model combined with a region and treatment/indication specific market demand pricing. In the context of specific oncology cancer indications in the US market, the integrated framework provides valuable economic viability assessment results coupled with the identification of key levers that influence product profitability within the cost model (batch, dose and donor yield) and market demand pricing model (standard of care, recurrence rate and survival rate). The proposed integrated framework enhances strategic decision making by providing early insights in the economic consequences of development- and commercialization scenarios to be used by pharmaceutical companies to bring life-saving innovative treatments to patients.

Student Presenters

Yuan Fang

Yuan is an electrical engineer with 17 years experience working in healthcare.

Ravi Kumar

With over 15 years of dynamic leadership in Engineering and Operations management, I bring a passion for leveraging technology to revolutionize business operations. My track record speaks volumes, having spearheaded initiatives that generated approximately $150M in profit. A seasoned... Read More →

Advisors

Jarrod Goentzel

Director, MIT Humanitarian Supply Chain Lab, MIT CTL

Friday May 17, 2024 1:30pm - 2:00pm EDT
Skyline AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

HEC-Healthcare

1:30pm EDT

Exploring Supply Chain Co-location: Implications on Cost, Speed-to-Market, and Sustainability

In today’s competitive business environment, companies are increasingly seeking ways to optimize their supply chain networks to reduce costs, improve responsiveness, and enhance sustainability. This capstone project explores the potential of co-location strategies, specifically the Supplier Park Model, in addressing supply chain inefficiencies for ABC Corporation, a leading food and beverage company. By comparing the baseline scenario of the company’s current geographically dispersed network with a future state scenario implementing the Supplier Park Model, the study quantifies the impact of co-location on three key metrics: cost reduction, speed-to-market improvement, and environmental sustainability. The results demonstrate that the Supplier Park Model can lead to a 45% reduction in total costs, a 30% improvement in speed-to-market time, and an 82% reduction in Scope 3 carbon emissions from transportation. The project highlights the importance of optimizing lead time, transportation, and inventory management in supply chain network design and provides actionable recommendations for ABC Corporation to implement the Supplier Park Model. This study contributes to the field of supply chain management by demonstrating the potential of co-location strategies to drive significant improvements in efficiency, responsiveness, and sustainability, pushing the boundaries of supply chain network design.

Student Presenters

Nilay Kumar

Student (MIT), ex-J&J, PwC

I am a dedicated supply chain professional currently pursuing my master's degree in Supply Chain Management at MIT. With a background in both Supply Chain Management at Johnson & Johnson and Supply Chain Consulting at PwC US Advisory, I've partnered with industry leaders to solve... Read More →

Oscar Nieto Michelis

Advisors

Eva Ponce

Director, Omnichannel Distribution Strategies, MIT Center for Transportation & Logistics

Friday May 17, 2024 1:30pm - 2:00pm EDT
Longfellow AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SCS-Supply Chain Strategy

1:30pm EDT

ELD Insights: Profiling Driver Duty Cycles and Truck Stop Visits in America

This capstone project presents a detailed profile of truck driver duty cycles and truck stop visits in the United States through the analysis of Electronic Logging Device (ELD) data. Addressing the gap in utilizing ELD-generated location data for broader analyses, this study evaluates drivers' unspent duty hours and stop durations at truck stops. Utilizing data from 1,261 long-haul drivers over 56 days, we employed Geospatial and Driver duty cycle analysis in Python to examine driver behaviors and truck stop patterns, revealing critical insights into duty cycle patterns and truck stop usage. Our findings indicate a high unspent duty hour per cycle, as evidenced by the analysis of over 36,000 duty cycles, highlighting that drivers spent an average of 9.9 hours out of the allocated 14 duty hours per shift, with considerable variability in work patterns. The stop durations at the truck stops vary depending on the purpose of the visit which was revealed in this project.

Student Presenters

Ardi Prasetya Hernawan

Advisors

David Correll

Research Scientist, MIT Center for Transportation & Logistics

Friday May 17, 2024 1:30pm - 2:00pm EDT
Longfellow C 40 Edwin Land Blvd, Cambridge MA 02142, USA

TRS-Transportation

2:00pm EDT

Unveiling Gross Profit Erosion Factors in a Medical Devices product Portfolio

A leading medical device company has experienced a consistent decline in profit margins across its product categories. Factors such as high inventory costs, complex Supply Chain operations, and challenges in managing a highly complex product portfolio have contributed to the erosion of gross profit. This study aims to uncover and address the factors eroding gross profit by exploring three key research questions: How can the company effectively segment its Endosurgery products to identify profit-eroding factors? What are the primary factors affecting the profitability of these products? Can unsupervised machine learning techniques, inspired by the BCG and Category Role matrices, provide a more comprehensive understanding of the factors influencing the business unit's financial performance compared to traditional approaches? The findings of this study contribute to the development of actionable strategies for enhancing product portfolio management and improving financial performance in the medical device industry. Within the product portfolio, 62% of the total product base codes have dilutive gross profit, driving further exploration of the factors influencing that behavior using Machine Learning techniques.

Student Presenters

Yaniliz Rivera

Amine Tmimi

SCM Residential Student, MIT Centre of Transporation and logistics

With more than seven years of experience in commodity management, strategic procurement, and supplier management, I am dedicated to harnessing the power of business analytics to enhance strategic sourcing decisions. My passion lies in understanding the crucial role that supply chains... Read More →

Advisors

Jim Rice

Deputy Director - MIT CTL, MIT Center for Transportation & Logistics

Jim Rice joined the MIT Center for Transportation & Logistics in 1995 and was appointed as the Deputy Director of CTL in 2007. In this capacity he oversees all industrial outreach programs and serves as the Director of the Supply Chain Exchange, including marketing and communication... Read More →

Friday May 17, 2024 2:00pm - 2:30pm EDT
Skyline AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

HEC-Healthcare

2:00pm EDT

Efficient Drop Trailer Management in a Volatile Network

This paper examines the implementation of a drop trailer program within a circular economy, focusing on optimizing the supply chain for a pallet manufacturing company. The study presents an integrated approach that combines an XGBoost machine learning model for precise demand forecasting and a Gurobi-based MILP optimization model for efficient trailer inventory management. By addressing key issues such as dwell time, stockouts, and operating hours, the research aims to enhance asset utilization and reduce operational costs. The quantitative model captures the movement of goods and the associated costs, informing strategic decisions about asset allocation, trailer scheduling, and cost minimization. The paper demonstrates that through advanced analytics and optimization techniques, substantial improvements in supply chain efficiency and cost savings can be realized, contributing to improved service levels and profitability.

Student Presenters

Alex Carroll

https://www.linkedin.com/in/alex-carroll629/

Troy Egar

Masters SCM Student ‘24, MIT

Advisors

Elenna Dugundji

Research Scientist, MIT Center for Transportation and Logistics

Thomas Koch

Postdoctoral Associate, MIT Center for Transportation and Logistics

Alex Carroll Troy Egar Executive Summary Review pdf

Friday May 17, 2024 2:00pm - 2:30pm EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

MAL-Machine Learning

2:00pm EDT

Transforming Supply Chain Strategy with Robotics: Measuring the Impact of Utilizing Robotics for Product Repackaging Operations

The US Bureau of Labor Statistics states that food and beverage manufacturers have experienced annual 0.5% decreases in labor productivity and annual 7% increases in unit labor costs since 2019 (US Bureau of Labor Statistics, 2023). These statistics underscore a growing inefficiency in the manufacturing and distribution processes of food products, posing significant challenges to industry players. Our sponsor, a Fortune 500 food manufacturer, illustrates these challenges in their product repackaging operations. Repackaging operations involve the meticulous case-packing of finished goods tailored to the specifications of end customers, typically retailers. The sponsor's product repackaging operation is costly, unscalable, and labor-intensive. Our research aims to assist our sponsor company by exploring the use of robotic systems to perform the product repackaging operation. Our methodology involved defining scope of the capstone, mapping processes, collecting and processing data, building a Mixed Integer Linear Programming (MILP) model and conducting scenario analysis to measure cost implications in various supply chain network design scenarios.

Student Presenters

Arjun Kapoor

Alex Varnhagen

Advisors

Chris Mejia

Director, MIT SCALE Network, MIT Food and Retail Operations Lab

Mauricio Gamez-Alban

Postdoctoral Associate, MIT Center for Transportation & Logistics

Kapoor Varnhagen Executive Summary pdf

Friday May 17, 2024 2:00pm - 2:30pm EDT
Longfellow AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SCS-Supply Chain Strategy

2:00pm EDT

Leveraging Freight Flow Data to Identify Underserved Industrial Real Estate Markets

This project aims to assist a logistics-focused real estate investment company in proactively identifying underserved markets in the U.S. transportation sector. Utilizing a mix of data from public and private sources and machine learning methods, the goal is to develop a quantitative methodology that highlights potential market investment opportunities for high flow-through (HFT) logistics facilities. The outcome includes a visualization tool to guide investment decisions and a market summary, enabling the company to capitalize on underserved logistics real estate markets.

Student Presenters

Santiago Hernandez

Shane Huisman

Advisors

Devadrita Nair

Postdoctoral Associate, MIT CTL

Ilya Jackson

Postdoctoral Associate, MIT Center for Transportation and Logistics

Friday May 17, 2024 2:00pm - 2:30pm EDT
Longfellow C 40 Edwin Land Blvd, Cambridge MA 02142, USA

TRS-Transportation

2:30pm EDT

Demand Forecasting with Machine Learning

Our capstone project focuses on forecasting sales of the sponsor company's Heat-not-Burn (HNB) products. We estimate future sales of consumables and kits in Italy by looking at monthly data between 2015 and 2023. Sales of the products have a solid positive trend, and regardless of any other parameter, we observe sales increase. We have received 117 features from the partner company, including but not limited to macroeconomic indicators, pricing of the sponsor company and competitors, and sales figures of the sponsor company and competitors. Our approach is first forecasting using traditional methods. After that, we apply different machine learning models. We compare the accuracies to see the difference between traditional and machine learning models. In addition to accuracy, we explore the explainability of the developed models. We use the SHAP algorithm to identify the features that contribute the most to the results.

Student Presenters

Jason Tseng

Turgay Turkmen

Advisors

Juan Carlos Pina Pardo

Postdoctoral Associate, Massachusetts Institute of Technology

Postdoctoral Associate at the MIT Megacity Logistics Lab. My research focuses on quantitative modeling to improve urban last-mile logistics.

Friday May 17, 2024 2:30pm - 3:00pm EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

MAL-Machine Learning

2:30pm EDT

Recurrent Neural Network for Predicting Sequential Supply Chain Delays

This capstone project addressed the challenges faced by GlaxoSmithKline’s (GSK) supply chain in managing sequential delays, essential for ensuring timely healthcare delivery in the pharmaceutical industry. The key objectives included pinpointing planned dates within GSK’s system and developing a robust machine learning model to predict sequential delays accurately. Through an extensive literature review and methodology development, the project focused on utilizing neural network machine learning methods, specifically recurrent neural networks (RNNs) and Long Short-Term Memory (LSTM) networks. Detailed summary statistics showcasing the delay frequencies and locations within GSK operations for the Benlysta brand revealed that approximately 40% of data exhibited delay issues, primarily in primary manufacturing sites. Further examination highlighted specific areas prone to delays, providing GSK with managerial insights for targeted action. The RNN model development involved data acquisition, manipulation, preprocessing, and model construction, followed by hyperparameter tuning to optimize performance, resulting in a reduced mean absolute error (MAE) of 4.89 days. Although challenges in linking manufacturing and quality data limited the initial scope, the project provided valuable insights and laid a solid foundation for future enhancements. Leveraging the findings and insights gained from this capstone project, GSK can enhance operational efficiency, mitigate supply chain risks and deliver medications to patients more effectively.

Student Presenters

Amber Lin

Anirudh Narula

Advisors

Tim Russell

Program Engineer, MIT Center for Transportation and Logistics

Tim Russell is a Program Engineer at the MIT Humanitarian Supply Chain Lab and the MIT CAVE Lab. Prior to this role, Mr. Russell conducted research with WFP on their cash and voucher program in Darfur. He has worked across the Caribbean, Latin America, Former Yugoslavia, and East... Read More →

Lin Narula Executive Summary final pdf

Friday May 17, 2024 2:30pm - 3:00pm EDT
Skyline AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

MAL-Machine Learning

2:30pm EDT

Behind the Curtain: Evaluating Supply Chain Responsiveness

In the dynamic beverage industry, rapid shifts in market demands necessitate an agile and efficient supply chain. Our sponsor, a CBA Beverage Manufacturer, aimed to assess its supply chain responsiveness to these ever-changing conditions. We leveraged key performance indicators, notably lead times, and employed both descriptive and predictive statistical techniques to analyze their internal process efficiency. Our analysis identified crucial predictors and assessed their impact, offering targeted strategies to address critical bottlenecks, particularly in the Southwest region, to excel in a fluctuating market landscape.

Student Presenters

Loreto Cantu

Graduate Student at MIT

Prior to MIT, I worked as a demand planner at OXXO, enhancing forecasting accuracy and inventory management. I hold a BS in Industrial & Systems Engineering from Tecnológico de Monterrey, Mexico.

Ray Zhang

Advisors

Inma Borrella

Research Scientist, MIT Center for Transportation and Logistics

Dr. Inma Borrella is a Research Scientist at the MIT Center for Transportation and Logistics (MIT CTL). She is the Academic Lead of the MITx MicroMaster’s in Supply Chain Management program, coordinating Massive Open Online Courses as well as the on-campus MIT Supply Chain Bootcamp... Read More →

Loreto & Ray Executive Summary pdf

Friday May 17, 2024 2:30pm - 3:00pm EDT
Longfellow AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

SCS-Supply Chain Strategy

2:30pm EDT

Equity and Profitability in Same-Day Delivery

The varying distances of customer locations from fulfillment centers and the diverse density of customer bases pose challenges in achieving service equity for Same-Day Delivery (SDD). Major SDD providers often prioritize operational profitability over equity across different neighborhoods in large cities. This capstone project aims to design a solution that assists SDD providers in preserving profitability while ensuring equitable planning of SDD routes. We propose dynamic policies that facilitate immediate decision-making regarding accepting or rejecting SDD orders. These policies are designed to help SDD providers grasp the implications of profitability by integrating equity factors into their operations. Using real-world data from the city of Vienna, our results indicate substantial enhancements in equity without significant compromise in profitability.

Student Presenters

Jean Herode Bertrand

Shino Yan Xu

Advisors

Juan Carlos Pina Pardo

Postdoctoral Associate, Massachusetts Institute of Technology

Postdoctoral Associate at the MIT Megacity Logistics Lab. My research focuses on quantitative modeling to improve urban last-mile logistics.

Matthias Winkenbach

Director of Research, Massachusetts Institute of Technology

Friday May 17, 2024 2:30pm - 3:00pm EDT
Longfellow C 40 Edwin Land Blvd, Cambridge MA 02142, USA

ULG-Urban Logistics

3:00pm EDT

Predicting semiconductor component lead time for an oil and gas company: A dynamic safety stock model with machine learning

In the oil and gas industry, the fluctuations of semiconductor component delivery time significantly impact Printed Circuit Board Assembly (PCBA) production planning. The discrepancies between the supplier quoted lead time and actual delivery lead time present a substantial challenge, as the sponsor company`s MRP system lacks a robust safety stock model to mitigate component shortage. To enhance the predictability of the delivery lead time and the standard deviation of lead time, this project introduces a machine learning-based framework. To mitigate semiconductor components shortage, this project also introduces a dynamic safety stock model, which incorporates the predicted delivery lead time and standard deviation of lead time. Through a comparative analysis of model performance, the tree model emerged as the most effective in predicating delivery lead time. The dynamic safety stock model also demonstrated improvements in inventory management and production planning. These improvements significantly reduce the risk associated with semiconductor supply chain variability and strengthen the company`s operation resilience.

Student Presenters

Chu Cho (Chong)

Advisors

Thomas Koch

Postdoctoral Associate, MIT Center for Transportation and Logistics

Chong Zhang Executive Summary pdf

Friday May 17, 2024 3:00pm - 3:30pm EDT
Skyline AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

PRC-Procurement

3:00pm EDT

Beyond Revenue: Comprehensive Risk Management in Global Warehouse Networks

Recent global events, particularly the COVID-19 outbreak, highlighted vulnerabilities in supply chains and the need for resilience in maintaining essential services, emphasizing the requirement for a robust Business Continuity Management (BCM) Plan. The sponsor (A leading pharmaceutical company with a global presence) currently has a revenue-first approach to risk management. Through this project, we aim to expand the sponsor’s current approach to include financial factors, inventory capabilities at each warehouse, customer profiles, and current business continuity management practices. This project's scope covers all warehouses within the sponsor’s downstream network, focusing on last-mile delivery from the sponsor’s warehouses to its patients. The deliverables include a framework through which the sponsor can understand the risk concentration for each of their warehouses depending on business needs. By providing a more holistic view of risks and incorporating lessons learned from the pandemic, the project aims to enhance the sponsor's ability to adapt to evolving challenges in the pharmaceutical supply chain, ensuring continuity and stability in the face of unforeseen disruptions.

Student Presenters

Nimisha Kunnathu

Linyi Zhang

Sr. Merchandising Manager, Spreetail

Advisors

Jafar Namdar

Postdoctoral Associate, MIT Center for Transportation and Logistics

Sreedevi Rajagopalan

Postdoctoral Associate, MIT Center for Transportation and Logistics

Friday May 17, 2024 3:00pm - 3:30pm EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

RMG-Risk Management

3:00pm EDT

Estimating On-Shelf Availability of CPG Products at Nanostores in India

Our sponsor (a multinational CPG company) faces the challenge of measuring On-Shelf Availability (OSA) for nanostores due a lack of inventory visibility at a store level. Nanostores, in emerging countries, are the channel that sells more than half of CPG products, performing a key role for these companies. Our sponsor company supplies 2.4 million nanostores in India, this approximately represents 20% of their sales in this country. OSA measures the quantity of products available for the customer to purchase at a specific moment. It is a key indicator to quantify product visibility for retailers, helping companies to identify stock-out events, and, most importantly, devise strategies to increase sales. The sponsor company provided us with distributor-level data, supplemented by a two-month field study currently underway. Our model's scope centers on one of India's largest regions, Mumbai, analyzing four nanostores sales channels, along with the top 20 selling SKUs within the region. Our methodology begins by shaping the demand for each key SKU using a stochastic distribution. We're considering either Poisson distribution, given the low quantities ordered, or Triangle distribution, due to the quantity of available data. We use the stochastic demand distribution in combination with the latest order point in time to find a probability of a stockout event from occurring. The model intends to create the ability to estimate OSA in situations where traditional methods fail due to the magnitude of stores and a lack of technological tracking systems.

Student Presenters

Matthias Eder

Stephanie Gomez

Advisors

Inma Borrella

Research Scientist, MIT Center for Transportation and Logistics

Friday May 17, 2024 3:00pm - 3:30pm EDT
Longfellow AB 40 Edwin Land Blvd, Cambridge MA 02142, USA

ROP-Retail Operations

3:00pm EDT

Elucidating US Import Supply Chain Dynamics: A Spatio-Temporal Graph Neural Network Approach

To enhance understanding of congestion points at ports and provide visibility into the incoming goods flow into the USA, this study focuses on maritime ports, using the Port of Boston and New York/New Jersey as case studies. Based on the Automatic Information System (AIS) data, we aim to develop predictive models for port congestion status and the Estimated Time of Arrival (ETA) of container ships. Additionally, we analyze historical commodity flow data to forecast future values, weights, volumes and categories based on Harmonized System (HS) codes. Employing quantitative AIS data analysis provides insights into port congestion dynamics and commodity flow trends, indicating the potential to improve the accuracy of ETA, port management and logistics visibility. This study contributes to both theoretical and practical applications in maritime logistics.

Student Presenters

Nikolay Aristov

Student, CTL MIT

I have more than 15 years of experience in implementing and maintaining SCM software. I performed technical roles on SCM projects, diving into different stages of supply chain planning and execution: demand planning, SRM, WMS, transportation, and master planning. Furthermore, I participated... Read More →

Ziyan Li

student, MIT

Advisors

Elenna Dugundji

Research Scientist, MIT Center for Transportation and Logistics

Thomas Koch

Postdoctoral Associate, MIT Center for Transportation and Logistics

SCM38 Aristov Li pptx pptx

ExecutiveSummary Aristov Li SCM38 pdf

Friday May 17, 2024 3:00pm - 3:30pm EDT
Longfellow C 40 Edwin Land Blvd, Cambridge MA 02142, USA

TRS-Transportation

3:30pm EDT

Forecasting the Demand of Key Opinion Leader (KOL) Live Streams in China

Key Opinion Leaders (KOLs) or influencers are known to significantly impact consumer demand, occasionally triggering unprecedented spikes. As a result, it has become increasingly important for companies to use KOLs to boost their sales and, more importantly, accurately forecast sales and effectively manage inventory levels. In collaboration with a leading company in the fast-moving consumer goods (FMCG) industry who utilizes KOLs to promote products during live streams, we aim to analyze the underlying features that influence the effectiveness of these events and build a model that forecasts future demand. We first start with qualitative analysis and then proceed with statistical analysis to describe the key features that drive sales for live stream events. The results of our study quantify the impact of KOLs on sales and provide insights demonstrating what features drive their effectiveness. Ultimately, this information will help our sponsor company refine their planning process, support top-line growth, and minimize excess inventory.

Student Presenters

Mavis Lu

Prior to MIT, I worked in denim and apparel sourcing and production at Ariat International. I holds CSCP credential by APICS and a BFA in Fashion Merchandising from Academy of Art University.

Elizabeth Marchosky

Student, Massachusetts Institute of Technology

Advisors

Jafar Namdar

Postdoctoral Associate, MIT Center for Transportation and Logistics

Executive Summary 2024 Lu Marchosky pdf

Friday May 17, 2024 3:30pm - 4:00pm EDT
Skyline CDE 40 Edwin Land Blvd, Cambridge MA 02142, USA

DMP-Demand Planning

3:30pm EDT

Improving Nutrition Rankings for Food Banks

Food insecurity is a major issue for many Americans. Food banks and food pantries strive to provide enough food to satisfy the needs of people suffering from food insecurity. These organizations are also trying to distribute healthy food, not just meeting the caloric intake necessary to sustain the community. The Mid-Ohio Food Collective (MOFC) is a food bank outside of Columbus, Ohio that was interested in measuring the healthiness of their inventory in a context that could be communicated to those outside of the hunger relief system. Specifically, they wanted to use a scoring system called the Healthy Eating Index (HEI), which is used by healthcare professionals. The healthcare industry wants to alleviate certain long term health issues, like diabetes, by teaming with food banks and other organizations to provide healthy food options as a preventative measure. HEI is a 0-100 measure of the nutritional quality of a set of food, with higher scores being healthier. Using a subset of the MOFC inventory, I used Microsoft Excel and Python to provide the food bank with an HEI score that they could communicate with healthcare providers to show they are distributing healthy food to their clients. The subset of the MOFC inventory had a score of 80.623 out of 100. While there is room for improvement, this was a promising start as MOFC can track scores and make changes to their purchasing decisions to raise scores over time. Though the process of producing the score is not fully automated, it is generalizable and can be performed for other entities in the hunger relief system, allowing them to demonstrate the healthfulness of the products they distribute as well.

Student Presenters