Taxi Trip Analysis Pipeline

Overview

This data wrangling project analyzes NYC taxi trip data, joins it with daily weather data, detects anomalous trips using a robust z-score method, and generates a deterministic dataset fingerprint.

I carried out this analysis using Python >=3.12,<3.13, Jupyter Notebook and managed with the uv Python package manager.

---

Environment Setup

Install dependencies:

uv sync

If uv is not installed:

pip install uv

---

Dataset Files

I placed the following datasets in the project root directory:

• yellow_tripdata_2022-07.parquet
• nycentral_city_weather_data.csv
• taxi_zone_lookup.csv

---

Weather Station ID

Station ID used for weather data: USW00094728

---

Running the Notebook

Navigate to the analysis.ipynb file and open it.

Then select a python kernel with a version >=3.12,<3.13.

Proceed to run all cells from top to bottom (or if available, click the "Run All" button) to reproduce the analysis and generate the outputs.

---

Outputs Generated

Running the notebook produces the following files inside the outputs/ directory:

outputs/task1_hourly_volume.png
outputs/task1_speed_hist.png
outputs/task1_summary.csv
outputs/task2_hhi.png
outputs/task2_top_routes.csv
outputs/task3_join_quality.csv
outputs/task3_rain_amount.png
outputs/task3_rain_duration.png
outputs/task4_anomalies.csv
outputs/task4_anomaly_scatter.png
outputs/task5_fingerprint.txt

---

Reproducibility

The fingerprint generation is deterministic. A seed is derived from the student ID and a salt using SHA256. Because the seed is fixed, the same dataset and code will always produce identical outputs when the notebook is executed.

Taxi Trip Analysis Pipeline

Overview

This data wrangling project analyzes NYC taxi trip data, joins it with daily weather data, detects anomalous trips using a robust z-score method, and generates a deterministic dataset fingerprint.

I carried out this analysis using Python >=3.12,<3.13, Jupyter Notebook and managed with the uv Python package manager.

---

Environment Setup

Install dependencies:

uv sync

If uv is not installed:

pip install uv

---

Dataset Files

I placed the following datasets in the project root directory:

• yellow_tripdata_2022-07.parquet
• nycentral_city_weather_data.csv
• taxi_zone_lookup.csv

---

Weather Station ID

Station ID used for weather data: USW00094728

---

Running the Notebook

Navigate to the analysis.ipynb file and open it.

Then select a python kernel with a version >=3.12,<3.13.

Proceed to run all cells from top to bottom (or if available, click the "Run All" button) to reproduce the analysis and generate the outputs.

---

Outputs Generated

Running the notebook produces the following files inside the outputs/ directory:

outputs/task1_hourly_volume.png
outputs/task1_speed_hist.png
outputs/task1_summary.csv
outputs/task2_hhi.png
outputs/task2_top_routes.csv
outputs/task3_join_quality.csv
outputs/task3_rain_amount.png
outputs/task3_rain_duration.png
outputs/task4_anomalies.csv
outputs/task4_anomaly_scatter.png
outputs/task5_fingerprint.txt

---

Reproducibility

The fingerprint generation is deterministic. A seed is derived from the student ID and a salt using SHA256. Because the seed is fixed, the same dataset and code will always produce identical outputs when the notebook is executed.