https://github.com/hiteshmeta85/ml-mini-project-api

Classification of Disaster related tweets from Social Media using BERT Model
https://github.com/hiteshmeta85/ml-mini-project-api

bert binary-classification flask machine-learning sqlite twint

Last synced: about 1 month ago
JSON representation

Classification of Disaster related tweets from Social Media using BERT Model

• Host: GitHub
• URL: https://github.com/hiteshmeta85/ml-mini-project-api
• Owner: hiteshmeta85
• Created: 2022-09-22T16:12:44.000Z (over 3 years ago)
• Default Branch: main
• Last Pushed: 2022-10-20T14:54:19.000Z (over 3 years ago)
• Last Synced: 2025-02-08T03:14:11.039Z (over 1 year ago)
• Topics: bert, binary-classification, flask, machine-learning, sqlite, twint
• Language: Jupyter Notebook
• Homepage:
• Size: 3.85 MB
• Stars: 0
• Watchers: 1
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# 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

   ![](images/index.png)

   

2. Enter Custom Hashtags

   ![](images/custom-hashtags.png)

3. Raw Data scraped from Twitter using Twint

   ![](images/raw-data.png)

4. Binary Classified Data using BERT Model

   ![](images/binary-classified-data.png)

5. BERT model flow for Binary Classification of Text

   ![](images/model-flow.png)