singapore-hawker-centers-model

Singapore Hawker Centers Status Prediction Model

Model Description

This is a machine learning model that predicts the status of Singapore hawker centers based on their characteristics. The model uses a Random Forest classifier to classify hawker centers into different status categories.

Model Performance

• Accuracy: 88.46%
• Algorithm: Random Forest Classifier
• Training Data: 129 Singapore hawker centers
• Features: 8 engineered features

Model Details

Model Architecture

• Type: Random Forest Classifier
• Number of Estimators: 100
• Random State: 42

Training Data

The model was trained on Singapore hawker centers data from the GeoJSON dataset, containing:

• 129 hawker centers
• Location coordinates (longitude, latitude)
• Number of food stalls
• Completion dates
• Status information

Feature Engineering

1. Age: Calculated from completion year
2. Location: Normalized longitude and latitude
3. Stalls: Number of cooked food stalls
4. Size Categories: Small (<20), Medium (20-50), Large (50-100), Very Large (>100)

Target Classes

• Existing: 103 centers (79.8%)
• Existing (new): 16 centers (12.4%)
• Under Construction: 6 centers (4.7%)
• Existing (replacement): 3 centers (2.3%)
• Other: 1 center (0.8%)

Feature Importance

1. Age: 55.2% - Most important predictor
2. Latitude: 20.6% - Geographic location matters
3. Longitude: 9.3% - East-west positioning
4. Stalls: 8.8% - Size of the center
5. Size Categories: 4.1% - Categorical size information

Usage

Python Example

import joblib
import numpy as np

# Load the model
model = joblib.load('hawker_centers_model.pkl')

# Prepare features for prediction
# [stalls, age, longitude_norm, latitude_norm, stall_Small, stall_Medium, stall_Large, stall_Very Large]
features = [50, 20, 0.5, 0.3, 0, 1, 0, 0]

# Make prediction
prediction = model.predict([features])
confidence = model.predict_proba([features]).max()

print(f"Predicted status: {prediction[0]}")
print(f"Confidence: {confidence:.2%}")

Input Format

• stalls: Number of food stalls (integer)
• age: Age of hawker center in years (integer)
• longitude_norm: Normalized longitude (0-1, float)
• latitude_norm: Normalized latitude (0-1, float)
• stall_Small: Binary indicator for small size (0 or 1)
• stall_Medium: Binary indicator for medium size (0 or 1)
• stall_Large: Binary indicator for large size (0 or 1)
• stall_Very Large: Binary indicator for very large size (0 or 1)

Training Details

Data Preprocessing

• Extracted completion years from date strings
• Calculated age from completion year
• Normalized coordinates to 0-1 range
• Created size categories based on stall count
• Grouped rare classes to avoid stratification issues

Model Training

• Train/Test Split: 80/20
• Cross-validation: Not used (small dataset)
• Class Balancing: Rare classes grouped as "Other"

Limitations

1. Small Dataset: Only 129 samples, which limits generalization
2. Class Imbalance: Most centers are "Existing" status
3. Geographic Bias: Model trained only on Singapore data
4. Temporal Bias: Data from specific time period
5. Feature Limitations: Only basic features used

Ethical Considerations

• Model should not be used for discriminatory purposes
• Results should be interpreted with caution due to small dataset
• Consider local context when applying to other regions
• Regular retraining recommended as new data becomes available

Citation

@misc{singapore-hawker-centers-model,
  title={Singapore Hawker Centers Status Prediction Model},
  author={Custom AI Studio},
  year={2024},
  url={https://huggingface.co/your-username/singapore-hawker-centers-model}
}

License

This model is released under the MIT License.

Contact

For questions or issues, please open an issue on the model repository.