How vishwanatha-14/ml2-mini works
This repository trains and compares five classifiers for four mobile-phone price ranges, then exposes saved models through a prediction CLI. The main workflow is specification data → cleaning and stratified split → model training and tuning → shared evaluation and saved artifacts → prediction for a new phone. The README also documents notebook-based EDA. Source excerpts support the training, evaluation, and prediction connections; notebook internals and artifact contents were not inspected.
Main entry points: train.py Train and tune · predict.py Prediction CLI · evaluate.py Metrics and plots · preprocessing.py Clean and split
Components 11
Data and EDA
- Training data
data/train.csv - Clean and split
src/preprocessing.py - Exploratory analysis
notebooks/mobile_price_classification.ipynb
Model training
- Train and tune
src/train.py - Saved classifiers
models
Evaluation and prediction
- Metrics and plots
src/evaluate.py - Comparison outputs
results - Prediction CLI
src/predict.py - Price range and probabilities
src/predict.py
Other components
- Analyst
- Phone specifications
Connections 13
- Train and tune → Training data: reads
src/train.py - Train and tune → Clean and split: cleans and splits
src/train.py - Train and tune → Metrics and plots: calls
src/train.py - Train and tune → Saved classifiers: saves
src/train.py - Train and tune → Comparison outputs: writes
src/train.py - Metrics and plots → Comparison outputs: writes reports
src/evaluate.py - Analyst → Train and tune: runs training
README.md - Analyst → Prediction CLI: requests prediction
README.md - Phone specifications → Prediction CLI: provides specs
src/predict.py - Prediction CLI → Saved classifiers: loads model
src/predict.py - Prediction CLI → Clean and split: uses labels and features
src/predict.py - Prediction CLI → Price range and probabilities: returns prediction
src/predict.py - Analyst → Exploratory analysis: explores data
README.md