Predicting Default Credit Card Clients

Project Description

Demonstrate how MLFlow works by using the Credit Card Default Dataset

Project Coverage

PART 1

Setup MLFlow Experiment for Manual Tuning
Create Runs for Manual Tuning Experiment (captures different parameters based on user input)
Save Experiments and Runs on local server

Save Experiments and Runs on a remote server (DagsHub)

PART 2

Setup MLFlow Experiment for HyperParameter Tuning
Create Runs for Hyperparameter Tuning Experiment

Run 1: DecisionTreeClassifier - Best Model
Run 2: DecisionTreeClassifier - Different Predictors
Run 3: DecisionTreeClassifier - Different Numerical Transformations
Run ∞: Repeat Runs using other classifier models

Model

Decision Tree

Data Source

https://archive.ics.uci.edu/static/public/350/default+of+credit+card+clients.zip

Installation

Install all requirements by running the following command

pip install -r requirements.txt

Project Configurations

Hyperparameter Tuning: Manual
Pipeline: NA
Model Tracking: MLFlow
Deployment: NA

Project Folder Structure

├── ...
├── 01_src  				# Source code
│   ├── download_data.py
├── 02_data
│   ├── 01_raw  			# Raw data files
│   ├── 02_processed 			# Processed data files
│   └── 03_external  			# Data from external sources
├── 03_notebooks  			# Notebooks used for pre-processing, exploration, model training, etc 
├── 03_src  				# Source code
├── 04_models  				# Trained model files, model metadata, and evaluation results
├── 05_deployment  			# Project deployment files
├── 06_reports  			# Project documentation, Jupyter Notebook reports, final reports, and presentations
├── 07_config  				# Configuration files for hyperparameters, data sources, logging, environment, database, and deployment
├── 08_tests 				# Unit tests or test scripts
├── 09_environments 		        # Environment setup file (dependencies)
├── README.md
└── ...

Pull Requests

If you have something to add or a new idea to implement, you are welcome to create a pull request on improvement.

References

• MLFlow Documentation
• Introduction to MLFlow
• Setting Up MLFlow Experiments to a Remote Server
• Kaggle Notebook
• MLFlow Reference Notebook 1
• MLFLow Reference Notebook 2

Give it a Star

If you find this repo useful, give it a star so as many people can get to know it.