Add Medical_Appointment_No_Shows: ML model

Author: farhann-saleem

projects/prediction/Medical_Appointment_No_Shows/data/KaggleV2-May-2016.csv

@@ -0,0 +1,102 @@
+# Medical Appointment No-Show Prediction
+
+## Description
+This project aims to predict patient no-shows for medical appointments using machine learning. No-shows cause inefficiencies and lost resources in healthcare; predicting these helps optimize scheduling and patient outreach.
+
+We explore several machine learning models, including Logistic Regression, Decision Trees, Random Forest, XGBoost, and Artificial Neural Networks (ANN). We perform data cleaning, feature engineering, model training, hyperparameter tuning, and evaluation.
+
+## Project Structure
+```
+Medical_Appointment_No_Shows/
+├── data/              # Dataset files (KaggleV2-May-2016.csv)
+├── notebooks/         # Jupyter notebooks (main.ipynb)
+├── src/              # Source code (Python modules)
+├── models/           # Saved models
+├── pictures/         # Visualizations and plots
+├── research/         # Research materials
+├── requirements.txt  # Dependencies
+└── README.md        # Project documentation
+```
+
+## Dataset
+**Source:** Kaggle Medical Appointment No-Shows Dataset
+
+The dataset contains patient appointment information, including:
+- **Patient demographics:** Age, scholarship status, health conditions (hypertension, diabetes, alcoholism, handicap)
+- **Appointment information:** Scheduled day, appointment day, days difference
+- **SMS reminders:** Whether patient received SMS reminder
+- **Target variable:** No-show labels (0 = showed up, 1 = no-show)
+
+**Key Challenges:**
+- Imbalanced dataset with fewer no-shows than shows
+- Time features needing careful processing (handling time components in dates)
+- Feature selection and encoding to improve model learning
+
+## Installation
+```bash
+# Clone the repository
+git clone <your-repo-url>
+cd Medical_Appointment_No_Shows
+
+# Install dependencies
+pip install -r requirements.txt
+```
+
+## Usage
+1. **Data Exploration:** Open `notebooks/main.ipynb` to explore the dataset
+2. **Data Preprocessing:** The notebook includes data cleaning and feature engineering
+3. **Model Training:** Train models using the provided code in the notebook
+4. **Evaluation:** Review model performance metrics and visualizations
+
+```python
+# Example: Load and explore data
+import pandas as pd
+df = pd.read_csv('data/KaggleV2-May-2016.csv')
+df.head()
+```
+
+## Model Details
+### Algorithms Used:
+- **Logistic Regression:** Baseline linear model
+- **Decision Tree:** Non-linear decision boundary model
+- **Random Forest:** Ensemble of decision trees with tuning
+- **Support Vector Machine (SVM):** Kernel-based classifier
+- **XGBoost:** Gradient boosting with class balancing
+
+### Performance:
+- **Baseline Accuracy:** ~77% (but poor recall on no-shows)
+- **Tuned Random Forest F1 Score:** ~0.44
+- **XGBoost Recall:** ~79% (improved no-show detection)
+
+### Training Details:
+1. Data Cleaning and Feature Engineering  
+2. Exploratory Data Analysis with visualization (histograms, bar plots, correlation heatmaps)  
+3. Baseline model comparisons (Logistic Regression, Decision Tree, Random Forest, SVM)  
+4. Hyperparameter tuning using GridSearchCV with custom scoring (F1)  
+5. Evaluation with classification metrics focusing on F1 score and recall  
+6. Advanced models: XGBoost with balanced class weights
+
+## Results
+- Baseline models achieved up to ~77% accuracy but poor recall on no-shows
+- Hyperparameter tuning improved F1 score significantly (~0.44), emphasizing recall
+- XGBoost model achieved ~79% recall for no-show detection
+- Final tuned models balanced recall and precision for effective no-show detection
+- Feature importance analysis revealed key predictors (SMS received, days difference, age)
+
+**Visualizations:** See `pictures/` folder for feature importance plots, confusion matrices, and other visualizations.
+
+## Contributing
+Contributions, suggestions, and feedback welcome! 
+
+**How to contribute:**
+1. Fork the repository
+2. Create a feature branch (`git checkout -b feature/improvement`)
+3. Commit your changes (`git commit -am 'Add new feature'`)
+4. Push to the branch (`git push origin feature/improvement`)
+5. Open a Pull Request
+
+## License
+MIT License
+
+---
+

projects/prediction/Medical_Appointment_No_Shows/git.ignore

@@ -0,0 +1,7 @@
+pandas>=1.5.0
+numpy>=1.23.0
+matplotlib>=3.6.0
+seaborn>=0.12.0
+scikit-learn>=1.2.0
+xgboost>=1.7.0
+jupyter>=1.0.0