This is a machine learning web application project (developed for my college lab practicals) that predicts outcomes based on the Apple Quality Prediction dataset. It uses Node.js with Express.js & EJS for the frontend and FastAPI for the backend, based on the kaggle notebook by Satvik. Developed together with Ishita Pradhan, who worked on the frontend while I worked on the backend and deployment.

• Frontend: Node.js, Express.js, EJS – Manages the UI.
• Styling: CSS – Custom UI design.

• Backend: Uvicorn, FastAPI (Python) – Handles ML model predictions and remote server.
• Machine Learning: Scikit-learn, NumPy, Pandas

✅ Train and evaluate six classification models
✅ Data normalization and scaling (L1, L2, Standard, MinMax)
✅ Hyperparameter tuning using GridSearchCV and 10-fold cross-validation
✅ Interactive web interface for making predictions

The dataset used is apple_quality.csv, containing the features:

• Size
• Weight
• Crunchiness
• Juiciness
• Ripeness
• Acidity
• Quality

The following six models were applied to classify apples into good or bad quality:

• K-Nearest Neighbors (KNN)
• Support Vector Machine (SVM)
• Decision Tree
• Random Forest
• Naïve Bayes
• Gradient Boosting

Each model was evaluated using the following metrics:

• Accuracy
• Precision
• Recall
• F1-Score

To improve model performance, the dataset was normalized and scaled using four different techniques:

• L1 Normalization
• L2 Normalization
• Standard Scaling
• MinMax Scaling

Each model was then re-evaluated after scaling, and the results were recorded.

To optimize model performance, GridSearchCV was used for hyperparameter tuning, along with 10-fold cross-validation to find the best model.

The models were evaluated again using:

• Cross-validated Accuracy
• Precision
• Recall
• F1-Score

The best model SVM with 92.075% Cross Validation Accuracy was selected based on these metrics, and then FastAPI was configured to run on the backend using the Uvicorn module, using the best model hyperparameters discovered in the notebook, developed by Satvik.

Node modules and frontend scripts with ejs and js were processed and developed by me.

git clone https://github.com/ravesandstorm/web-app-apple-quality.git  
cd web-app-apple-quality

pip install -r requirements.txt

cd frontend  
npm install

cd backend
uvicorn main:app --reload

npm run dev

The application will be available at http://localhost:3000.

If you'd like to contribute, feel free to fork the repository and submit a pull request.

This project is licensed under the MIT License.