ML-MINI-PROJECT-API

Dataset

https://www.kaggle.com/competitions/nlp-getting-started/data?select=train.csv

Accuracy achieved: 80.44%

collab Notebook link Training - https://colab.research.google.com/drive/1NoBShinYljfw_sEKyF0TEDzGdlfaZRsT?usp=sharing

collab Notebook link Inference - https://colab.research.google.com/drive/1ThpOIpY33l3WXFGH9QAlX0BmE67pTZNK?usp=sharing

Setup [Docker]

Build Docker Image

docker-compose up

Rebuild Docker Image

docker-compose up --build

Setup [Local]

Model Used - BERT UNCASED

Download and save the folder in backend/models/

Model link - https://drive.google.com/file/d/1IJBNMt2pGmDxTq2e64rpzDjaz4aAauL7/view?usp=sharing

Installation

• Twint installation

pip3 install --user --upgrade git+https://github.com/twintproject/twint.git@origin/master#egg=twint

• Install requirements

pip install -r requirement.txt 

Backend - FLASK

If you want to start a new session od db

• Delete the db.sqlite file in backend folder
• Run the following commands

python init_db.py

cd backend
python app.py

---

Sample Images

1. Select Disaster Type

   

2. Enter Custom Hashtags

   

3. Raw Data scraped from Twitter using Twint

   

4. Binary Classified Data using BERT Model

   

5. BERT model flow for Binary Classification of Text