NYC Taxi Trip Data Analysis

Welcome to my NYC Taxi Trip Data Analysis project! This repository showcases the work I have done as a junior data scientist to analyze and model New York City taxi trip data for January 2025. Below, I explain what I did step by step and summarize the results of my analysis.

---

Table of Contents

0. Import libraries
1. Import data
2. Data visualization
3. Data cleaning
4. Data preparation
5. Benchmark model
   5.1. Train-test split
   
6. Data preparation

---

What I Did

1. Imported Libraries

Back to top

I started by importing essential Python libraries such as:

• pandas for data manipulation
• numpy for numerical operations
• matplotlib and seaborn for data visualization
• scikit-learn for machine learning modeling

import numpy as np import matplotlib.pyplot as plt import pandas as pd import seaborn as sns

2. Loaded the Dataset

Back to top

I worked with the NYC Yellow Taxi trip dataset for January 2025, which was provided in .parquet format. The dataset contains detailed information about taxi trips, including:

• Pickup and drop-off times
• Trip distances
• Passenger counts
• Fare details (e.g., fare amount, tips, total amount)

3. Data Exploration and Visualization

Back to top

I explored the dataset to understand its structure and identify key patterns:

• Visualized distributions of trip distances, fare amounts, and passenger counts.
• Analyzed trends over time (e.g., hourly, daily patterns).
• Identified outliers such as negative fares or unusually high values.

4. Data Cleaning

Back to top

To ensure data quality, I performed the following steps:

• Removed rows with missing or invalid values (e.g., negative fares).
• Filtered out extreme outliers that could skew the analysis.
• Ensured consistency in categorical variables like payment types.

5.1. Train-test split

Back to top

• Split the data into training and testing sets to evaluate model performance.

6. Built a Benchmark Model

Back to top

I trained a simple machine learning model to predict taxi fare amounts based on trip distance, time of day, and other features:

• Used a DecisionTreeRegressor as the benchmark model.

---

Results

Insights from Data Analysis

1. Trip Patterns:

   • Most trips occurred during rush hours (morning and evening).
   • Weekends had fewer trips compared to weekdays.

2. Fare Trends:

   • Fare amounts were generally proportional to trip distances.
   • Short trips within Manhattan had higher average fares per mile compared to longer trips.

3. Impact of Weather:

   • Rainy days showed a slight increase in trip fares due to longer travel times.

Model Performance

The benchmark model achieved the following results:

• Mean Absolute Error (MAE): ~3.50 USD
  This indicates that the model's predictions were off by an average of $3.50 compared to actual fare amounts.

---

How to Use This Project

1. Clone this repository:

git clone https://github.com/your-repo-name/nyc-taxi-analysis.git


cd nyc-taxi-analysis

2. Install required dependencies:

pip install pandas numpy matplotlib seaborn scikit-learn

3. Open the Jupyter Notebook (notebook_taxi_nyc.ipynb) to explore the analysis and results step by step.

What I Learned

This project helped me build skills in:

1. Cleaning and preparing real-world datasets.
2. Performing exploratory data analysis (EDA) using visualizations.
3. Building and evaluating machine learning models.
4. Understanding how external factors (e.g., weather) impact predictions.