Predicting Student's Math Scores

Check it out : Students Math Score

Overview

This repository contains code for building a machine learning model to predict students' math scores based on various features such as gender, race/ethnicity, parental level of education, lunch type, and test preparation course.

The aim is to develop an accurate model, focusing on writing production-level code and creating data pipelines from data acquisition to preprocessing and predicting.

Dataset

The dataset used for this project consists of the following columns:

• Gender
• Race/Ethnicity
• Parental Level of Education
• Lunch Type
• Test Preparation Course
• Math Score (target variable)
• Reading Score
• Writing Score

Machine Learning Models Utilized

Several machine learning algorithms were explored to develop the predictive model. The models used include:

• Linear Regression
• Ridge Regression
• Lasso Regression
• Support Vector Regression (SVR)
• Decision Tree Regression
• Random Forest Regression
• K-Nearest Neighbors Regression
• Gradient Boosting Regression
• AdaBoost Regression
• CatBoost Regression
• XGBoost Regression

Model Evaluation and Selection

To identify the best performing models, techniques such as Randomized Search Cross-Validation (RandomizedSearchCV) were employed to tune hyperparameters and optimize model performance. Model evaluation metrics such as mean squared error (MSE), root mean squared error (RMSE), and R-squared values were utilized to assess predictive accuracy and generalization capabilities.

Final Model Selection

After thorough evaluation, the best-performing models were selected based on their predictive accuracy and performance metrics. These models were potentially combined or further fine-tuned to create the best final predictive model for estimating students' math scores based on the given features.

Summary

This project entails developing a comprehensive data processing and modeling pipeline utilizing Python, Flask, Docker, and AWS for deployment. Tasks include conducting thorough EDA, feature engineering, model training, website creation, Dockerization, CI/CD implementation, and AWS deployment setup. The objective is to deliver a robust, scalable solution for data analysis and predictive analytics.

Project Structure

1. Setup GitHub and Local Folder

   • Create GitHub repo and .gitignore
   • Create venv
   • Create setup.py
   • Create requirements.txt

2. Create Source Code Structure

   • Create src directory and build the package (requirements.txt)
     • Create component files: data_ingestion.py, data_transformation.py, model_trainer.py
     • Create pipeline files: predict_pipeline.py, train_pipeline.py
     • Create exception, logger, and utils files: exceptions.py, logger.py, utils.py

3. Exploratory Data Analysis (EDA) in Jupyter Notebook

   • Perform EDA
   • Handle missing values
   • Remove duplicate values
   • Data cleaning
   • Data imputation
   • Feature engineering
   • Train-test split
   • Identify best performing models
   • Model evaluation (R2)

4. Create Simple Webpage for User Input

5. Write Modular Code with respect to the Jupyter Notebook and Test on Local Server (Flask)

6. Docker Configuration and Deployment

   • Docker setup and configuration

   sudo apt-get update -y
   sudo apt-get upgrade
   
   curl -fsSL https://get.docker.com -o get-docker.sh
   
   sudo sh get-docker.sh
   
   sudo usermod -aG docker ubuntu
   
   newgrp docker
   

   • Build Docker image

7. Configure GitHub Workflow and CI/CD Action Runner

8. Setup AWS Resources for Deployment

   • Create and configure IAM user

   • Set up Amazon ECR repository

   • Provision EC2 instance for deployment

Final Result