guillaumelgf / nlp-experiments

Simple NLP project

NLP-for-fun

Simple NLP project

To Do:

• Extract data from Kaggle
• Use python script 1_json_to_csv_dataset to convert json to csv readable by pandas
• Create ML Model to predict article category from title
  • Remove stop words (NLTK)
  • Stemming (NLTK)
  • Bag of words
  • Random Forest (scikit-learn)
• Serving (Flask)
  • Basic
  • A/B Testing models
  • Load new models

Logs:

• [20/01/2021]: First quick model
  • Convert data from json to csv
  • Implementation of quick feature Engineering (bag of word)
  • DecisionTreeModel not tuned
  • First evaluation

BUG: Vocabulary list needs to be made before splitting

How to make it work ?

Installing

git clone https://github.com/guillaumelgf/NLP-for-fun.git
cd NLP-for-fun
pip -m venv venv
.\venv\Scripts\activate
pip install -r requirements.txt

Training

python train.py

Starting flask app

set FLASK_APP=flask_app.py
set FLASK_ENV=devlopment
flask run

Then go to localhost:5000