Language Classifier

This project focuses on training an algorithm to correctly identify the language that the user types in.

The algorithms I used were:

• K Nearest Neighbors
• Multinomial Naive Bayes
• Random Forest

Out of these three algorithms, Random Forest performed the best (although Multinomial Nayes Bayes was a close second).

Note: In case you're wondering how I chose the parameters for each algorithm: I used RandomizedSearchCV from the sci-kit learn library to arrive at those parameters. I did not include the code for this because it took a long time to run each time.

How to Run the Application

There are two ways to run the Streamlit web application:

1. Visit the website

The simplest way to view the web app is by visiting the following link: https://share.streamlit.io/johng034/language-classifier/app.py

2. Run locally

If you wish to run the application on your machine, then complete the following steps:

1. Clone the repository (click here for instructions on how to clone a GitHub repository)
2. Open the folder of this repository in your editor of choice (or in the terminal/command prompt)
3. In the terminal, install the packages with pip install requirements.txt (you may want to install using a virtual environment for this)
4. Once the packages are installed, you can run the streamlit application by typing streamlit run app.py in the terminal

Improvements

I am currently considering adding data from wikipedia pages or tweets to train the algorithm on a wider range of data.