Logistic-Regression

A sorting hat for Hogwart's Houses.

A multi-class classifier, using logistic regression coded from scratch.

Final Score 125/100

Getting Started

First clone this repo.
git clone https://github.com/anyashuka/Logistic-Regression.git; cd Logistic-Regression

Download dependencies.
pip install -r requirements.txt

Then simply run the run_all.sh script to follow the following steps in each folder.
./run_all.sh

1.analyze

describe.py is Pandas describe() recoded from scratch, using Numpy.

Run with dataset_train.csv as argument, describe.py displays info on numerical feaures for dataset_train.csv.
python3 describe.py ../data/dataset_train.csv

The output can be compared with the output of Pandas describe().
python3 pandas_describe.py ../data/dataset_train.csv

2.visualize

Histogram

Which Hogwarts course has a homogeneous score distribution between all four houses? Arithmancy & Care of Magical Creatures

python3 histogram.py ../data/dataset_train.csv

Scatter_plot

What are the two features that are similar? Astronomy & Defense Against the Dark Arts

python3 scatter_plot.py ../data/dataset_train.csv

Pair_plot

What features are you going to use for your logistic regression?

python3 pair_plot.py ../data/dataset_train.csv

3.logreg

Train

Train model to predict Hogwart's House using logistic regression and dataset_train.csv. Save model to weights.csv.

python3 logreg_train.py ../data/dataset_train.csv -t -c

Flags

• -t, --timer. Display time taken to train. Takes nearly a minute on my system for 100000 epochs.

• -c, --cost. Display cost graph, prediction error over training period.

Predict

Predict houses for dataset_test.csv, using the model saved as weights.csv. Saves predicted houses to houses.csv.

python3 logreg_predict.py ../data/dataset_test.csv ../data/weights.csv

Test Accuracy

Test accuracy of predicted houses against dataset_truth.csv. In tools/ run:

python3 accuracy.py ../data/dataset_truth.csv ../data/houses.csv

Dependencies

Thankfully, running the following command should take care of dependencies for you.

pip install -r requirements.txt

Python 3.9.1

• pandas
• numpy
• seaborn
• termcolor
• matplotlib
• scikit_learn

Team

I wrote this project in a team with the awesome @dfinnis.