Tipping behavior in urban transportation is shaped by a complex mix of social, economic, and temporal factors. This project explores tipping patterns in New York City's taxi and for-hire vehicle services using 2023 trip data provided by the Taxi & Limousine Commission (TLC), supplemented with neighborhood-level median income data.
Our goal is to understand whether and how tipping is influenced by the socioeconomic context of pickup and dropoff neighborhoods. We examine trip-level attributes and geographic income disparities across NYC to predict tip amounts and uncover patterns of urban mobility compensation.
- Dependent variable: Tip amount or tip percentage.\
- Independent variables: Trip distance, fare, tip percentage, hour, day of week, season, pickup and dropoff locations, neighborhood median income, taxi type.
- Covers yellow taxi, green taxi, for hire vehicles trips across New York City during 2023.\
- Publicly available through the NYC Open Data platform.\
- Each row represents a single trip with details including pickup/dropoff times and locations, fare amount, tip amount, and payment type.
- Table: B19001 -- Household Income in the Past 12 Months (2023)\
- Retrieved from the U.S. Census Bureau's ACS 1-Year Estimates\
- Used to extract median income at the neighborhood (PUMA) level.\
- Income data was spatially joined to pickup/dropoff zones based on TLC taxi zones.
After cleaning and preprocessing, we used trip data from four months in 2023 (as specified in our code). Each processed dataset contains:
- Approximately 21 columns, including engineered features such as tip percentage and time-of-day variables.\
- The datasets vary in size depending on the month but each contains hundreds of thousands of rows.\
- Trips with zero fare, zero tip, or missing location data were excluded.\
- Geographic income data was merged using spatial joins to match taxi zones to median income estimates.
The following Python libraries are required to run the analysis and modeling process:
import numpy as np
import pandas as pd
import statsmodels.api as sm
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score
from sklearn.linear_model import LinearRegression
from sklearn.ensemble import RandomForestRegressor
import geopandas as gpd To predict tipping behavior in NYC taxi rides, we implemented supervised machine learning models---specifically, Linear Regression and Random Forest Regression. Linear Regression provides a straightforward and interpretable baseline for understanding the relationship between tipping and variables like fare, distance, and neighborhood income. It helps identify whether a linear relationship exists between predictors and the tip amount or percentage.
Random Forest is a more flexible, non-linear model that can capture complex interactions between features. It is particularly well-suited for our dataset, which includes both numerical and categorical variables, and helps address potential overfitting through ensemble learning. Using both models allows us to compare performance and better understand the underlying patterns in tipping behavior.
We evaluate model performance using Root Mean Squared Error (RMSE) and R² Score to assess accuracy and explanatory power. These metrics help determine how well the models predict tipping based on socioeconomic and trip-related features.
| Feature | Description | Type |
|---|---|---|
| Distance | Total distance of the trip (in miles) | Numerical |
| Fare | Total fare amount for the trip | Numerical |
| Tip Percentage | Tip as a percentage of the fare | Numerical |
| Hour | Hour of the day when the trip started | Categorical |
| Day of Week | Day of the week (e.g., Monday, Tuesday) | Categorical |
| Pickup Location | TLC taxi zone where the trip started | Categorical |
| Dropoff Location | TLC taxi zone where the trip ended | Categorical |
| Neighborhood Income | Median household income of the pickup/dropoff neighborhood | Numerical |
| Taxi Type | Type of taxi (Yellow, Green, Uber, Lyft) | Categorical |
To associate socioeconomic context with taxi trips, income data was merged with trip records by matching location identifiers to neighborhood-level income estimates. After cleaning and filtering the data to remove outliers and missing values, models were trained on a subset of trip data. Exploratory Data Analysis (EDA) was conducted using visualization libraries such as Matplotlib and Seaborn, allowing the team to identify trends, detect anomalies, and better understand the relationships between variables.
- NYC Taxi & Limousine Commission: https://www.nyc.gov/site/tlc/about/tlc-trip-record-data.page\
- U.S. Census Bureau, American Community Survey: https://data.census.gov/table/ACSDT1Y2023.B19001?q=B19001+&g=040XX00US36$7950000,36$9700000