Bluebikes-Data-Analysis

🚲 Bluebikes Boston (Jan–Sep 2025): Weather-Aware Demand, Forecasting & Rebalancing 🌦️

Course: BA775 – Fall 2025, Section A06
Notebook: A06-Bluebikes-Weather-Forecasting.ipynb
Team: Emily Su · Vishesh Goyal · Grace Kung · Mokhinur Talibzhanova · Anton Falk

---

📌 Project Overview

Boston’s Bluebikes system sees big swings in ridership across seasons, hours of the day, and especially weather conditions.
This project combines Bluebikes trip data (Jan–Sep 2025) with hourly weather data to:

• Understand how stations, routes, and riders behave under different weather and time-of-day conditions.
• Build weather-aware demand indicators to support bike rebalancing and reduce the risk of empty/full stations.
• Provide actionable recommendations for operations teams (where to add capacity, when to move bikes, and which stations are weather-resilient).

The core analysis is implemented in a single, well-documented notebook:
A06-Bluebikes-Weather-Forecasting.ipynb.

---

🔍 Key Questions We Answer

The notebook is structured around a set of guiding questions, including (but not limited to):

1. Which stations have the highest number of departures and returns?
2. What are the most popular start/end station routes?
3. How do peak usage hours differ, and how does ridership vary across months?
4. How does average trip duration differ between members and casual riders, and what does that imply for availability?
5. How does weather (temperature, rain, wind) influence total trips and demand patterns?
6. How does weather influence AM vs. PM peak usage differently?
7. Does wind speed actually slow riders (distance/duration), or just reduce total demand?
8. Which stations have the most stable demand regardless of weather (weather-resilient “backbone” stations)?
9. Which stations consistently lose bikes in the morning and gain them back in the evening (rebalancing hotspots)?
10. How elastic is demand with respect to rain and temperature (% change in trips per unit change in weather)?

Each question is backed by BigQuery SQL, visualizations, and a short “So what?” interpretation explaining the operational impact.

---

🧱 Data Sources

All data is accessed via Google BigQuery (schema and column descriptions are documented in the notebook).

1. Bluebikes Trip Data (Jan–Sep 2025)

• Monthly tables under a dataset similar to:
  ba775-fall25-a06.bluebike_weather.2025MM-bluebikes-tripdata
• Key columns include:
  • date – ride date
  • ride_id – unique ride identifier
  • rideable_type – classic vs. electric
  • started_at, ended_at – start and end timestamps
  • start_station_name, start_station_id
  • end_station_name, end_station_id
  • start_lat, start_lng, end_lat, end_lng
  • member_casual – rider type (member or casual)

2. Weather Data (Hourly)

• Joined to trips at the date / approximate hour level.
• Typical fields:
  • tempmax, tempmin, temp
  • precip, precipprob
  • windspeed, windgust
  • conditions / derived weather category

The notebook includes a table dictionary describing each field and the merged dataset schema.

---

📊 Dashboards & Visual Storytelling

The analysis is designed to support interactive dashboards (built in Tableau) that answer:

• Q1 & Q2: Station demand & route popularity
• Q3 & Q4: Trip duration vs. rider type, hourly & monthly peaks
• Q5–Q10: Weather-sensitive demand, elasticities, and station resilience

You can link your Tableau Public dashboards here, e.g.:
https://public.tableau.com/...

The notebook also embeds screenshots of final dashboards and stories for reference.

---

🧮 Analytical Approach

High-level workflow inside the notebook:

1. Data Engineering & Merging

   • Load Bluebikes monthly trip tables from BigQuery.
   • Join with hourly weather data into a single Bluebikes + Weather fact table.
   • Standardize station identifiers and ensure consistent date/time handling.

2. Exploratory Data Analysis (EDA)

   • Station-level demand (departures vs. returns).
   • Route-level flows (A→B vs. B→A comparisons).
   • Member vs. casual behavior (trip duration, days of week, seasonality).
   • Demand over time (hour of day, day of week, month).

3. Weather-Aware Metrics & Indices

   • Compare normal vs. rain vs. extreme heat days.
   • Derive weather sensitivity for each station (how much demand drops under bad weather).
   • Examine wind vs. speed (distance/duration) to see whether riders slow down or simply ride less.
   • Build elasticity views such as:

     % change in trips per inch of rain / per degree of temperature change.

4. Rebalancing Insights

   • Identify stations that consistently lose bikes in the morning and gain them back in the evening.
   • Highlight weather-resilient stations that keep demand stable even when conditions change.
   • Suggest rebalancing strategies:
     • Prioritize high-impact central stations.
     • Use weather sensitivity to plan truck routes and bike redistribution.
     • Focus on corridors where under-supply or over-supply is most common.

---

🗂️ Repository Structure

Suggested structure for this repo:

.
├── A06-Bluebikes-Weather-Forecasting.ipynb   # Main analysis & narrative
├── README.md                                 # You are here 🙂
├── img/                                      # Exported dashboard screenshots (optional)
│   ├── dashboard1.png
│   └── dashboard2.png
└── sql/                                      # Optional: saved BigQuery SQL scripts
    ├── station_demand.sql
    ├── weather_join.sql
    └── elasticity_metrics.sql