Course Project: Machine Learning for Chemical Engineering

Open In Colab

Course Project: Machine Learning for Chemical Engineering#

Weight: 50% of final grade

Overview#

Apply machine learning techniques to a chemical engineering (or your major if you are not a chemical engineer) problem of your choice. This project demonstrates your ability to:

  1. Formulate a meaningful problem

  2. Collect and preprocess data

  3. Apply appropriate ML methods

  4. Evaluate and interpret results

  5. Communicate findings clearly

Project Options#

Option A: Bring Your Own Data#

Use data from your research, internship, or a published paper. This is the preferred option as it connects to your own work.

Option B: Public Dataset#

Use a publicly available dataset relevant to chemical engineering:

Example Topics#

  • Predicting material properties from composition

  • Process optimization using surrogate models

  • Fault detection in manufacturing processes

  • Reaction yield prediction

  • Catalyst performance modeling

  • Polymer property prediction

Deliverables#

1. Project Proposal#

  • Problem description (1 paragraph)

  • Data source and description

  • Proposed methods

  • Expected outcomes

2. Progress Report / Check-in#

  • Data exploration and preprocessing

  • Initial modeling results

  • Challenges encountered

  • Updated plan

3. Final Report#

  • Complete analysis notebook

  • Written report (see template below)

4. Presentation#

  • 10-minute presentation

  • 5-minute Q&A

Report Template#

Your final report should include:

1. Introduction (1-2 pages)#

  • Problem motivation

  • Background and prior work

  • Objectives

2. Data (1-2 pages)#

  • Data source and collection

  • Feature descriptions

  • Exploratory data analysis

  • Preprocessing steps

3. Methods (2-3 pages)#

  • Model selection rationale

  • Hyperparameter tuning approach

  • Validation strategy

4. Results (2-3 pages)#

  • Model performance metrics

  • Comparison of methods

  • Feature importance / interpretability

  • Uncertainty quantification (if applicable)

5. Discussion (1-2 pages)#

  • Key findings

  • Limitations

  • Future work

6. Conclusions (0.5 page)#

  • Summary of contributions

Grading Rubric#

Criterion

Description

Problem Formulation

Clear, relevant, well-motivated

Data Analysis

Thorough EDA, appropriate preprocessing

Methodology

Appropriate methods, proper validation

Results

Clear presentation, proper metrics

Interpretation

Domain insights, meaningful conclusions

Communication

Clear writing, good visualizations

Tips for Success#

  1. Start early - Data collection and cleaning take time

  2. Iterate - Don’t expect perfect results on the first try

  3. Document everything - Keep notes on what you tried

  4. Validate results - Check if findings make physical sense

  5. Ask questions - Office hours are there to help

  6. Focus on insight - A simple model with good interpretation beats a complex black box

Contents