data-portfolio

Taxi Trip Analysis — NYC Yellow Taxi (2017)

Exploratory data analysis of NYC yellow-taxi trips (2017) to profile trips, identify fare/tip drivers, and surface operations levers for pricing & cashless strategy.

Project Overview

Title: NYC Taxi Trips — Fare & Operations Insights
Business problem: Which factors most influence total fare (distance, duration, passengers, payment method, tips)? What does a typical trip look like? Which actions (pricing, cashless/tip nudges) can improve revenue and operational KPIs?
Dataset: Taxi_Trip_Data.csv — 22,699 rows × 18 columns (NYC TPEP schema), period 2017-01-01 → 2017-12-31.
Tools used: Python (Pandas, NumPy, Matplotlib), Jupyter Notebook.

Analysis Results

Key finding 1 — Trip profile & fare drivers
- Median distance ≈ 1.61 mi; median duration ≈ 11.18 min.
- Median total_amount ≈ $11.80 (p90 ≈ $30.35).
- Correlations with total_amount: fare_amount 0.984, trip_distance 0.902, dur_min 0.162, passenger_count 0.011 → distance/fare dominate; duration/passengers contribute little.
- Median fare_per_mile ≈ $5.51 (robust to outliers).
Key finding 2 — Payment & tipping behavior
- Payment mix: Card 67.29%, Cash 32.04% (others <1%).
- Tip% (card): median ≈ 22% (mean ≈ 21.4%).
- ~71% of trips have 1 passenger → clear target for cashless/tipping nudges.
Business impact
- Pricing: calibrate by distance (not duration); monitor fare per mile.
- Revenue uplift: promote cashless (higher/trackable tips); set default tip buttons at 20–25%.
- Ops KPIs: focus on tip rate and fare per mile rather than passenger count.

Dashboard & Files

Adjust relative paths to your repo layout.

Notebook: Taxi-Trip-Analysis.ipynb
Dataset (CSV): Taxi_Trip_Data.csv

Technical Highlights

Size: 22,699 trips (raw); 22,681 after cleaning.
Summary: distance median 1.61 mi; duration median 11.18 min; total_amount median $11.80, p90 $30.35.
Tip (card): median ~22%, mean ~21.4%.
Correlations to total_amount: fare 0.984, distance 0.902, duration 0.162, passengers 0.011.
Derived metric: fare_per_mile median $5.51.

Skills Demonstrated

Data wrangling (Pandas), datetime feature engineering, outlier handling, correlation analysis, and decision-ready business.