Project Summary
Taxi Trip is one of the major transportation modes especially in big cities, such as New York City. The taxi industry in New York City is controlled by Taxi Limousine Commission (TLC).
Taxi Trip is a flexible type of transport, where passengers or hirers choose where and when their trip starts and ends.
Demand for taxis varies over time. So, unveiling the factors that affect demand and understanding the dynamics of taxi demand will not only benefit the driver, but many other groups too. It helps managements to improve their transportation systems, it helps drivers to reduce their vacant time and more importantly, it helps drivers to earn more.
Project Overview: The project aimed to build machine learning models to accurately predict taxi fares based on trip data in New York City. Utilizing Python’s data science and machine learning libraries, the project focused on providing accurate fare estimates to enhance operational efficiency, improve customer satisfaction, and optimize taxi service pricing.
- DELIVERABLES
- Utilized Python to develop a machine learning model to predict taxi trip fares, leveraging advanced algorithms to provide accurate fare estimates based on trip details and conditions.
- Collected and preprocessed a comprehensive dataset of green taxi trips, including features such as pickup and drop-off locations, time of day, distance traveled, and traffic conditions.
- Utilized feature engineering techniques to extract relevant predictors, such as time of day, day of the week, and weather conditions, enhancing the model’s accuracy and relevance.
- Implemented hyperparameter tuning and cross-validation techniques to optimize model performance, achieving high accuracy and minimizing prediction errors.
- Applied various machine learning algorithms, including linear regression,Random Forests, Support Vector and Gradient Boosting, to determine the best-performing model for fare prediction.
- Visualized the correlation between trip fares and other variables using Heatmap.
- Developed a robust evaluation framework using metrics such as Mean Squared Error (RMSE), and R-squared to assess model effectiveness.
- Conducted extensive geospatial analysis on taxi trip data to uncover patterns, trends, and insights related to trip distribution, hotspots, and spatial behavior.
- Collected and integrated geospatial features, including taxi trip coordinates, pickup and drop-off locations, and time stamps, to build a comprehensive spatial analysis framework.
- Utilized Geocoders library to get the Postcode of the pickup and dropoff cordinates, identifying key patterns and taxi trip distance.
- Utilized Pgeocode library to get the pickup and dropoff addresses, using their postcodes.
- Performed spatial clustering analysis to identify distinct geographic zones with varying levels of taxi activity, such as high-demand areas, underserved regions, high-paying areas and time.
- ANALYSIS IMPACT
- Successfully delivered a machine learning model capable of predicting taxi fares with high accuracy. The real-time fare estimation system will enhance the operational efficiency of taxi services and improve customer experience by providing reliable fare estimates. Detailed analysis and reporting offered actionable insights for optimizing fare pricing strategies and integrating the predictive models into operational systems.
- Reduce drivers’ free time by enabling them to position at the right addresses at every given time, enabling them to always have a passenger.
- Help drivers to earn more, since they already know the locations that pays more.
- Help taxi users to always have rides available for them.