NYC taxi demand in SQL

#Brief

A fleet operations lead asks you ten questions about January 2025 demand. They will use your answers to plan driver shifts at airports and in Manhattan. Use DuckDB (free, no server) directly on the TLC Parquet file and zone lookup.

#What to build

Answer, with SQL:

1. Trips per day, and the busiest and quietest days.

2. Trips by hour of day, weekday versus weekend.

3. Top 10 pickup zones with names and boroughs.

4. The most common pickup and dropoff zone pairs.

5. Median fare and median trip time from JFK and LaGuardia to Manhattan.

6. Share of trips paid by card, per borough.

7. Average tip as a share of fare for card payments, by hour.

8. Zones with no pickups in the month (anti-join).

9. Day-over-day and same-weekday-last-week change in trips.

10. The top 2 pickup zones in each borough (window functions).

Before question 1, run a data quality checklist and decide what to exclude.

#Deliverables

• answers.sql (or a notebook) with each question as a heading

• Results tables and one sentence of interpretation for each

• A data quality section: checks, counts, decisions

#How it is graded

Reviewers rerun your SQL. Correctness and explicit handling of bad rows carry the most weight; readable SQL is next. Wrong joins that inflate counts are the most common reason for low marks.

#Stretch

Repeat for February 2025 with union all and show which answers changed.