Project Assignment

Description

The course includes a semester-long project completed in teams of up to three students. Each team is responsible for selecting and analyzing a meaningful real-world problem using machine learning techniques.

Projects should resemble a small applied research or analytics study rather than simply applying models to data. Emphasis is placed on problem formulation, data-driven analysis, evaluation, and actionable insights.

Project Scope

  • Identify a real-world, impactful problem
  • Use machine learning as a tool (not the goal itself)
  • Focus on industrial engineering applications such as:
    • Operations
    • Supply chains
    • Healthcare systems
    • Energy systems
    • Transportation
    • Socio-technical systems

Data Requirements

  • Must include a data-driven experimental component
  • Prefer real-world datasets
  • Synthetic data allowed only with strong justification
  • Must include:
    • Data preprocessing
    • Modeling decisions
    • Evaluation and interpretation

Analysis Expectations

Projects must:

  • Clearly articulate the problem context
  • Justify modeling choices
  • Perform meaningful evaluation
  • Provide insights and recommendations
  • Discuss limitations and future improvements

Course Requirements & Grading

Grading Components

  • Project Proposal (25%)
  • Final Report (65%)
  • Reproducibility & Submission Quality (10%)

Grading Criteria

Projects will be evaluated based on:

  • Importance and clarity of the problem
  • Methodological soundness
  • Data usage and evaluation quality
  • Depth of analysis
  • Clarity of communication
  • Reproducibility

Course Materials

Content

  • Instructor-provided guidelines and expectations

A strong project:

  • Starts from a real decision or system problem
  • Is data-driven and reproducible
  • Includes careful evaluation
  • Produces stakeholder-relevant insights

Communication

  • Coordinate within teams professionally
  • Contact TA for logistics or grading questions
  • Use course platforms (Canvas/Ed) for clarifications

Policies, Expectations, & Guidelines

Project Policies

  • Teams must be self-formed
  • All members receive the same grade
  • Instructional staff will not arbitrate team disputes
  • Teams are responsible for collaboration and workload management

What Makes a Good Project

Strong Examples

  • Predicting emergency room overcrowding
  • Identifying supplier delays
  • Customer churn prediction
  • Detecting abnormal system behavior

Weak Examples

  • Applying a single algorithm without context
  • Comparing models without a real problem
  • Generic “analyze a dataset” projects

Project Deliverables

Deliverable 1: Project Proposal (Mid-Semester)

Required Sections

  1. Problem Statement
    • Real-world system and stakeholders
    • Why the problem matters
  2. Research Questions / Objectives
    • Clear prediction or analysis goals
  3. Dataset Description
    • Source and structure
    • Features and target variable
    • Dataset size
  4. Proposed Methods
    • At least one baseline model
    • Additional comparison models
    • Justification of choices
  5. Evaluation Plan
    • Metrics (e.g., accuracy, RMSE, F1)
    • Validation strategy
    • Definition of success
  6. Action Plan
    • Timeline and responsibilities
    • Key milestones

Deliverable 2: Final Project Report

Submission Package

Submit a zipped folder including:

  • Report (PDF)
  • Code (notebooks or scripts)
  • Data or access instructions
  • README for reproducibility
  1. Problem Statement and Goal
  2. Data Description
  3. Methods
  4. Evaluation Strategy
  5. Results
  6. Interpretation and Recommendations
  7. Limitations and Future Work

Technical Expectations

  • Include a baseline and at least one advanced model
  • Use proper evaluation (no training-set evaluation)
  • Provide tables/figures for results
  • Include basic error analysis