README.md

Product Category ML Classification Web App Service (ML-as-service)

The developed model has achieved the highest test accuracy of 99.812 % and 0.998 test F1-score.

The notebook codetemplate/notebooks/EDA.ipynb contains the data exploration and analysis. The modeling, experiments, training, and evaluation of the algorithms are explained in the codetemplate/notebooks/Modeling-Training-Evaluation.ipynb notebook.

The webapp is built on Flask framework which takes user input on the go and returns a predicted category and predicted probability.

codetemplate directory contains the source files, the unittests, the flask webapp, ipynb notebooks & web html, results for the project.

Model	Training Accuracy	Validation Accuracy	Test Accuracy	Test Precision	Test Recall	Test F1-Score	Training F1-score	Validation F1-score
Logistic Regression	99.969	99.755	99.625	0.99629	0.996	0.996	1.0	0.998
Multinomial Naive Bayes	100	99.802	99.688	0.996877	0.997	0.997	1.0	0.998
Perceptron	100	99.656	99.688	0.996895	0.997	0.997	1.0	0.997
CART	100.000	98.296	99.125	0.991449	0.991	0.991	1.000	0.983
Random Forest	100.000	99.380	99.500	0.995099	0.995	0.995	1.000	0.994
MLP	99.969	99.812	99.812	0.998127	0.998	0.998	1.0	0.998

Instructions:

1. Create a virtual environment such as conda or virtualenv

conda create --name productClass python=3.6

2. Install python dependencies using pip from the requirements.txt file

conda activate productClass
pip install -r requirements.txt

3. Run the unittests in the operations/tests/ in IDE to check if everything is alright. Make sure to define the exact project root path in the webapp/app.py for `operation/tests/unit/test_api.py'.
4. Run the train.py file to run the experiments and save trained models and the result files.
5. Start the flask web app by running the webapp/app.py.