Academic Success Classification Model

Overview

Welcome to the 2024 Kaggle Playground Series! This repository showcases the end-to-end pipeline for predicting academic risk in higher education students, leveraging a clean and comprehensive machine learning approach.

This project aims to classify students into categories (Graduate, Dropout, or Enrolled) based on their academic, demographic, and socioeconomic data.

---

Goal

To predict the academic risk of students in higher education and submit predictions with the highest possible accuracy, evaluated using the accuracy score.

---

Repository Structure

The repository is organized into the following directories:

1. Project_Documents

Contains textual documents describing the dataset, columns, and project overview:

• Column_names.txt: Describes the columns in the dataset.
• Dataset_Description.txt: Provides details about the dataset.
• ProjectOverview.txt: Summarizes the project goals and objectives.

2. Data Files

Contains the raw dataset and submission template:

• train.csv: Training dataset with labeled data.
• test.csv: Test dataset without labels.
• sample_submission.csv: Submission format example.

3. Understanding_Data

Contains Jupyter notebooks and documents for understanding the dataset:

• Academic_Success_UnderstandingData.ipynb: Notebook exploring the dataset structure.
• Academic_Success_UnderstandingData.pdf: PDF version of the notebook.
• Analysis of Training and Test Data.docx: Detailed analysis summary of the dataset.

4. Exploratory_Data_Analysis

Contains visualizations and insights derived during EDA:

• Academic_Success_EDA.ipynb: Notebook showcasing exploratory data analysis.
• Academic_Success_EDA.pdf: PDF version of the EDA notebook.
• Exploratory Data Analysis for Academic Success Classification Model.docx: Comprehensive EDA analysis document.

5. Predictive_Modeling

Contains the predictive modeling workflow, results, and submissions:

• Academic_Success_Classification_Model.ipynb: Notebook implementing CatBoost, XGBoost, and HistGradientBoostingClassifier.
• Academic_Success_Classification_Model.pdf: PDF version of the notebook.
• Predictive Modeling Analysis for Academic Success Classification Model.docx: Detailed analysis of model performances and insights.
• submission.csv: Final submission file with predicted labels.

---

Evaluation

Submissions are evaluated using the accuracy score. For each id row in the test set, the predicted Target (categorical academic risk assessment) is required. The submission format must include a header:

id,Target
76518,Graduate
76519,Graduate
76520,Graduate

---

Technologies and Tools Used

• Python Libraries: Pandas, NumPy, Seaborn, Matplotlib, Scikit-learn, CatBoost, XGBoost
• Data Visualization: Seaborn and Matplotlib
• Machine Learning Models: CatBoost, XGBoost, and HistGradientBoostingClassifier
• Jupyter Notebooks: Used for data exploration, EDA, and modeling
• Documentation: Microsoft Word for detailed explanations

---

Key Steps in the Pipeline

1. Understanding Data:

   • Dataset overview and distribution analysis.
   • Observed class imbalance and ensured no missing values.

2. Exploratory Data Analysis:

   • Visualized trends and relationships in features such as Gender, Course, Grades, Parental Occupation, GDP, and Inflation Rate.
   • Derived insights to support feature engineering and predictive modeling.

3. Predictive Modeling:

   • Implemented CatBoost, XGBoost, and HistGradientBoostingClassifier.
   • Evaluated models on validation data using accuracy, precision, recall, and F1-score.
   • Selected CatBoost as the best-performing model with an accuracy of 83.11%.

4. Submission:

   • Generated predictions for the test set using the selected model.
   • Prepared the final submission.csv file.

---

Instructions for Use

1. Clone this repository:

   git clone https://github.com/your_username/Academic_Success_Classification_Model.git

2. Navigate to the directory:

   cd Academic_Success_Classification_Model

3. Install dependencies:

   pip install -r requirements.txt

4. Explore the notebooks:

   • Open Understanding_Data/Academic_Success_UnderstandingData.ipynb for dataset analysis.
   • Open Exploratory_Data_Analysis/Academic_Success_EDA.ipynb for EDA insights.
   • Open Predictive_Modeling/Academic_Success_Classification_Model.ipynb for modeling workflows.

5. Generate predictions:

   • Run the Predictive_Modeling/Academic_Success_Classification_Model.ipynb notebook.
   • Save predictions in submission.csv format.

---

Conclusion

This repository provides a comprehensive framework for analyzing academic risk in higher education. The approach integrates data understanding, EDA, and state-of-the-art machine learning techniques to deliver actionable insights and accurate predictions.

Feel free to explore the repository, and contributions are welcome to enhance this project further!

---

Author

Javeriya Shaikh

• linkedin: [www.linkedin.com/in/javeriya-shaikh-57869a231]
• GitHub: [https://github.com/Shaikh-Javeriya]

For any questions or suggestions, feel free to reach out!