Urban Bike Demand Forecasting Model
A trained Random Forest regression model for predicting hourly bike rental demand using temporal, weather, seasonal, and operational features.
Model Overview
This model was developed as part of the Urban Demand Forecasting project using the Seoul Bike Sharing Demand dataset.
The model predicts the expected number of rented bikes for a given hour based on factors such as:
- Temperature
- Humidity
- Wind speed
- Visibility
- Dew point temperature
- Solar radiation
- Rainfall
- Snowfall
- Hour
- Season
- Holiday
- Functioning day
- Temporal features derived from the date
Model Details
| Property | Value |
|---|---|
| Model | Random Forest Regressor |
| Task | Regression |
| Framework | scikit-learn |
| Estimators | 200 |
| Random State | 42 |
| Preprocessing | ColumnTransformer |
| Categorical Encoding | OneHotEncoder |
| Unknown Categories | Ignored |
The preprocessing and model are stored together in the serialized .pkl file as a scikit-learn Pipeline.
Dataset
The model was trained using the Seoul Bike Sharing Demand dataset.
The dataset contains hourly bike rental demand along with weather and operational information from Seoul.
Input Features
The production pipeline expects the original input features and automatically derives additional temporal features from the Date column, including:
- Year
- Month
- Day
- Day of Week
- Weekend indicator
The Date column is used for feature engineering and is not directly passed to the final model.
Usage
The model can be loaded using joblib:
import joblib
model = joblib.load("bike_demand_model.pkl")
prediction = model.predict(input_data)
Intended Use
This model is intended for:
Urban mobility analysis
Bike rental demand forecasting
Machine learning experimentation
Predictive analytics demonstrations
Streamlit-based prediction applications
Limitations
Predictions depend on the quality and distribution of the input data.
The model was trained on the Seoul Bike Sharing Demand dataset and may not generalize directly to other cities.
Extreme weather or unusual events may produce less reliable predictions.
This model is intended for demonstration and predictive analytics purposes rather than operational decision-making without further validation.
GitHub:
https://github.com/Pranav123221/urban-demand-prediction-Forecasting-engine
Author
Pranav Sharma
GitHub: https://github.com/Pranav123221
LinkedIn: https://www.linkedin.com/in/pranav-sharma-333b67338/