Detective Dog

Twitter Profile Sentiment Analyzer
View Demo · Report Bug · Request Feature

📖 Table of Contents

Table of Contents

• About The Project
• Screenshots
• Tech Stack
• Models
• API Reference
  • Get Sentiment using Username
• Roadmap
• Run Locally
• Environment Variables
• Authors
• Contributing
• License
• Acknowledgements
• Related

About the Project

This Good Boy sniffs out the sentiment from Twitter profile and presents it in an elegant way.
Or you could also say it as Applying Sentiment Analysis on Twitter Profile and using ChartJs to visualize the result.

Screenshots

Tech Stack

Client:

• React
• Redux
• Axios
• ChartJs
• Notyf

Server:

• NLTK
• TextBlob
• VaderSentiment
• Sklearn
• Tweepy
• Twitter API
• Flask Restful API
• BeautifulSoup

Models

DataSet Used for Training : Sentiment140

Model		Training		Testing		API
Name	Settings	Accuracy	F1 Score	Accuracy	F1 Score	Version
Textblob	Default	-	-	62.24%	0.5994	v1
VaderSentiment	Default	-	-	65.19%	0.6453	v2
Logistic Regression	Count Vect. & Lemmatization Used	79.58%	0.8008	77.92%	0.7859	v3-1
	TF-IDF & Lemmatization Used	79.80%	0.8015	78.11%	0.7856	v3-2
Bernoulli Naive Bayes	Count Vect. & Lemmatization Used	80.36%	0.8051	76.98%	0.7733	v4-1
	TF-IDF & Lemmatization Used	80.36%	0.8052	76.98%	0.7733	v4-2
Multinomial Naive Bayes	Count Vect. & Lemmatization Used	80.37%	0.8020	76.91%	0.7680	v4-3
	TF-IDF & Lemmatization Used	80.36%	0.8051	76.98%	0.7733	v4-4
Gradient Boosting Classifier	TF-IDF { min_df=5 } & Lemmatization Used . Gradient Boosting Parameters : {lr=1.5, n=150, depth=10}	80.03%	0.8065	77.00%	0.7780	v5-1
	TF-IDF { min_df=5 } & Lemmatization Used . Gradient Boosting Parameters : {lr=1.25, n=300, depth=15}	83.72%	0.8410	76.89%	0.7747

Model Results

Following are the score for the username "elonmusk"

Model	API	Score
Name	Version	Positive Score	Negative Score
Textblob	v1	61.69%	38.31%
VaderSentiment	v2	69.71%	32.09%
Logistic Regression	v3-1	72.06%	27.94%
	v3-2	74.07%	25.93%
Bernoulli Naive Bayes	v4-1	68.40%	31.60%
	v4-2	72.21%	27.79%
Multinomial Naive Bayes	v4-3	67.44%	32.56%
	v4-4	62.47%	37.43%
Gradient Boosting Classifier	v5-1	64.86%	35.14%

API Reference

Get Sentiment using Username

  GET /sentiment-v{x}/<string:username>

{x} is the version number which hosts different ML models. Check Models section for more details

Parameter	Type	Description
username	string	Required. Twitter Profile Username

Roadmap

• Frontend using React
• Backend Model hosting on heroku using Flask Restful API
• Creating Sentiment model using TextBlob & Vader
• Custom ML Models :
  • Logistic Regression
  • Naive Bayes
    • Multinomial
    • Bernoulli
  • Gradient Boosting Classifier
• Custom DL Models :
  • CNN
  • RNN, LSTM
  • BERT
• Refactoring the WebApp Code
• Update Notebooks
• Improve Existing Models

Run Locally

Clone the project

  git clone https://github.com/utsavk28/DetectiveDog

Install dependencies in server folder.

  cd server
  python -m venv my-project-env
  source my-project-env/bin/activate
  pip install -r requirements.txt

Generate environment variables and fill in the values.

  cp .env.example .env

Your .env is ignored by git, which you can see in .gitignore, and so, it's safe!

Starting Development Server

  python main.py

Install dependencies in client folder.

  cd ../client # If you are in ./server
  npm i

  change API url in ./client/src/utils/api.js to your need

Starting Client

  npm start

At the end of this, you should have

• server running at http://127.0.0.1:5000/
• new_client running at http://localhost:3000/

Environment Variables

To run this project, you will need to add the following environment variables to your .env file

API_KEY : Twitter API/Consumer Key

API_KEY_SECRET : Twitter API/Consumer Secret

BEARER_TOKEN : Twitter Bearer Token

ACCESS_TOKEN : Twitter Access Token

ACCESS_TOKEN_SECRET : Twitter Access Secret

Authors

• Utsavk28

Contributing

Contributions are what make the open source community such an amazing place to learn, inspire, and create. Any contributions you make are greatly appreciated.

If you have a suggestion that would make this better, please fork the repo and create a pull request. You can also simply open an issue with the tag "enhancement". Don't forget to give the project a star! Thanks again!

1. Fork the Project
2. Create your Feature Branch (git checkout -b feature/AmazingFeature)
3. Commit your Changes (git commit -m 'Add some AmazingFeature')
4. Push to the Branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

License

Distributed under the MIT License. See LICENSE.txt for more information.

Acknowledgements

• Pre-Processing :
  • Analytics Vidhya Blog
  • Kaggle NLP 101 Preprocessing
  • Medium NLP Preprocessing Blog
• Machine-Learning-Projects Repo -> Twitter Sentiment Analysis
• ML-ProjectKart -> IMDB Sentiment Analysis
• PaperSpace Gradient Boosting Blog
• Flask Restful API Intro. & Deployment Blog
• Folder Structure Blog
• Regex Visualizer
• Regex CheatSheet
• Word Embedding Word2Vec
• StackEdit

Further Reading

• Sentiment Analysis using 7 techniques Kaggle
• Twitter Sentiment Analysis using Word2Vec Word Embedding
• Deep Convolutional Neural Network for Sentiment Analysis

Related

Here are some related projects

• Machine-Learning-Projects
• ML-ProjectKart