|
|
---
|
|
|
title: RidePricingInsightEngine
|
|
|
emoji: 🚗
|
|
|
colorFrom: blue
|
|
|
colorTo: green
|
|
|
sdk: gradio
|
|
|
sdk_version: "5.22.0"
|
|
|
app_file: app.py
|
|
|
pinned: false
|
|
|
license: mit
|
|
|
short_description: ML-powered ride fare prediction using regression
|
|
|
---
|
|
|
|
|
|
# Dynamic Ride Pricing Optimization
|
|
|
|
|
|
## 🚀 Overview
|
|
|
|
|
|
This project presents a machine learning-based solution for **ride fare prediction and optimization**, using a rich synthetic dataset of historical ride data. By analyzing real-time supply-demand conditions and customer attributes, the system aims to help ride-sharing platforms implement **data-driven dynamic pricing strategies**.
|
|
|
|
|
|
## 📊 Dataset
|
|
|
|
|
|
- **Source**: [Dynamic Pricing Dataset (Kaggle)](https://www.kaggle.com/datasets/arashnic/dynamic-pricing-dataset)
|
|
|
- **Description**: A **synthetic dataset** designed for ride-sharing fare prediction. Features include:
|
|
|
- Number of Riders / Drivers
|
|
|
- Location Category
|
|
|
- Loyalty Status
|
|
|
- Number of Past Rides
|
|
|
- Ratings
|
|
|
- Booking Time
|
|
|
- Vehicle Type
|
|
|
- Expected Ride Duration
|
|
|
- Historical Cost of Ride
|
|
|
|
|
|
## 🔍 Problem Statement
|
|
|
|
|
|
The objective is to identify which features most influence ride fares and approximate pricing behavior through various **regression models**. This includes:
|
|
|
- Finding key predictors for optimal fare setting.
|
|
|
- Estimating price sensitivity across rider and supply characteristics.
|
|
|
- Offering recommendations for **dynamic fare adjustments** in real time.
|
|
|
|
|
|
## 🧠 Model & Features
|
|
|
|
|
|
- **Regression-based modeling** (e.g., Linear, Ridge, XGBoost)
|
|
|
- **Final model**: Huber Regressor with non-zero intercept — robust to outliers, encouraging sparse feature usage
|
|
|
- **Feature selection pipeline** to reduce unnecessary variables
|
|
|
- **Price approximation engine** that generalizes across different booking conditions
|
|
|
- **Gradio-powered UI** for hands-on exploration
|
|
|
|
|
|
## 🛠️ Tools and Libraries
|
|
|
|
|
|
- **Python**
|
|
|
- **Scikit-learn**, **XGBoost**
|
|
|
- **Pandas**, **NumPy**
|
|
|
- **Gradio** for app deployment
|
|
|
- **Plotly** for interactive visualization
|
|
|
|
|
|
## 🧪 How to Run Locally
|
|
|
|
|
|
```bash
|
|
|
git clone https://github.com/Sharma-Pranav/Portfolio.git
|
|
|
cd Portfolio/projects/DynamicPricingOptimization
|
|
|
pip install -r requirements.txt
|
|
|
python app.py
|
|
|
```
|
|
|
|
|
|
## 📈 Results
|
|
|
|
|
|
- Clear insights into **fare-driving features** like rider demand, loyalty, and time of day.
|
|
|
- **Huber Regressor** outperformed others due to robustness and minimal feature reliance.
|
|
|
- **Optimal pricing zones** visualized for strategic recommendations.
|
|
|
- Portable deployment for teams to simulate and iterate pricing strategies.
|
|
|
|
|
|
---
|
|
|
|
|
|
> ✨ Developed by **Pranav Sharma** | 🚀 Hugging Face Space: [`RidePricingInsightEngine`](https://huggingface.co/spaces/PranavSharma/RidePricingInsightEngine)
|
|
|
|
