https://github.com/amajji/multi-class-classification

Deployment of a classification model on a webapp using FLASK for the backend and html/CSS/JS for frontend
https://github.com/amajji/multi-class-classification

analyse-data app classification data flask flask-application imbala imbalanced-classes imbalanced-classification imbalanced-data machine-learning machine-learning-algorithms preprocessing webapp webapplication

Last synced: about 1 hour ago
JSON representation

Deployment of a classification model on a webapp using FLASK for the backend and html/CSS/JS for frontend

• Host: GitHub
• URL: https://github.com/amajji/multi-class-classification
• Owner: amajji
• License: gpl-3.0
• Created: 2022-04-29T20:11:11.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2023-04-21T13:26:00.000Z (over 1 year ago)
• Last Synced: 2024-11-15T10:48:21.648Z (2 months ago)
• Topics: analyse-data, app, classification, data, flask, flask-application, imbala, imbalanced-classes, imbalanced-classification, imbalanced-data, machine-learning, machine-learning-algorithms, preprocessing, webapp, webapplication
• Language: Jupyter Notebook
• Homepage:
• Size: 23.1 MB
• Stars: 4
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Multiclass classification.

Data scientist | [Anass MAJJI](https://www.linkedin.com/in/anass-majji-729773157/)

***

## :monocle_face: Description

- This project aims to implement a multi-class classification model. we have four classes with a minority class (less than 1%), 

so we are in the case of unbalanced classes. We used regularization methods in order to penalize the errors made on this minority class.

The model is trained to predict around 4 different class. 

 

## :rocket: Repository Structure

The repository contains the following files & directories:

- **Dataset directory:** It contains a data pre-processing notebook where the train.csv file is used for training 

the model. Il contains also the predictions of test.csv dataframe.

- **model_weights:** It contains all the weights of the models : one-hot-encoder, target encoder, random forest model.

- **App directory:** Code for the web application that was developed for the model deployment. It contains Flask API code for the Back-End,

and HTML/CSS/Javascript code for the Front-End.

![](last_gif.gif)

## :chart_with_upwards_trend: Performance & results

- The test dataset contains **25 000 samples**. Each sample contains many features, and its corresponding label.

- The model used for this multi-class classification task is a **Random Forest** model.

- The metric used to measure the model's performance is **F1-score**. After testing the model, I obtained a test F1-score of **72 %**

---

## :mailbox_closed: Contact

For any information, feedback or questions, please [contact me][anass-email]

[anass-email]: mailto:[email protected]