This project aims to predict student dropout using machine learning techniques, involving data analysis, preprocessing, feature selection, model training, and evaluation. The dataset contains information about students, including demographics, academic performance, and other relevant factors. The code uses Streamlit to create a web interface for the dropout prediction model, loading a pre-trained Random Forest Classifier model and a Label Encoder for preprocessing categorical variables. Users can input student details and select options for categorical variables, with the model predicting whether the student is likely to drop out or not based on the input.
To get started with the project, you'll need Python, Streamlit and the following libraries installed:
- pandas
- numpy
- matplotlib
- seaborn
- scikit-learn
- imbalanced-learn
- xgboost
- yellowbrick
You can install these libraries using pip:
pip install pandas numpy matplotlib seaborn scikit-learn imbalanced-learn xgboost yellowbrick-
Clone the repository
-
Navigate to the project directory:
cd student-dropout-prediction- Run the Jupyter notebook to see the analysis, preprocessing, model training, and evaluation.
The dataset contains 395 records with 31 features. It includes information about students such as school, sex, age, family size, academic performance, etc.
- Logistic Regression
- Naive Bayes (BernoulliNB)
- K Nearest Neighbors
- Decision Tree
- Support Vector Machine
- Random Forest
- XGBoost
The Random Forest Classifier performed the best with an accuracy of 81% and an F1 score of 0.84 (approximately).
The project successfully predicted student dropout using machine learning. The Random Forest Classifier showed the best performance among the models tested with 81% accuracy.