This project walks through a machine learning life cycle from data exploration to model development and deployment by building a LGBTQ Friendly Tweet Classifier. This process is no mean linear but iterative.

graph LR
A((Data Exploration))-->B((Model Development))
B --> C((Deployment))
C --> D((Monitoring))
D --> A

Disclaimer: The accuracy of the deployment model needs a lot more improvement. I hope you don't use the result of the deployment model to harm the community but to iterate on it to make better models and projects.

@xingvoong

The dataset and details about it can be found here.

The final product is an LGBTQ tweet classifier that takes in a tweet to classify whether it is LGBTQ-friendly. The thought process for developing the model can be found in Notebooks.

Flow 1

Flow 2

There are 4 notebooks for this project. They explain the decision-making process through each phrase of an ML life cycle. At the end of each notebook, there is a summary section that concludes the takeaways of each notebook or each phrase.

While the project succeeded in showcasing an ML life cycle, it has a lot of limitations and room for improvement. Some of which I can think of:

• Same users are more likely to use the same languages for their tweets, therefor splitting data by users could decrease biases.
• Right now, the final product only classifies whether a tweet is friendly. It could do more by making a guideline or suggestion on how to write friendly tweets.
• Adding a feature to let users rate whether a prediction is good or bad, then taking that new data to train better models.

Clone the project

git clone https://github.com/xingvoong/ml-life-cycle-with-lqbtq-friendly-tweet-classifier/tree/main/notebooks

Go to the project directory

cd  ml-life-cycle-with-lqbtq-friendly-tweet-classifier

Creating and activating virtual environment

python3 -m venv env
source env/bin/activate

Installing needed packages

pip3 install -r requirements.txt

To run the prototype Flask app

At the project root directory, run:

python3 app.py

This will start the Flask development server, and you can access your application by navigating to http://127.0.0.1:5000/ in your web browser.

This project is made possible via the resources I am using and inspired by:

• LGBT Tweets 🏳️‍🌈 dataset from Kaggle
• Hate-Speech-Detection-in-Social-Media-using-Python by @NakulLakhotia
• The book Building ML Powered Applications by
  Emmanuel Ameisen