🚲 Bike Sharing Datasets

This repository provides tools to download, process, and structure bike-sharing datasets from multiple cities for analysis and research purposes.

📥 Datasets Download

You can download bike-sharing datasets from the following sources:

• New York: CitiBike
• Washington: Capital Bikeshare
• Bay Area: BayWheels
• Columbus: CoGo
• Boston: Blue Bikes

⚙️ Installation

Clone the repository and install dependencies:

git clone <repo_link>
cd <repo_directory>
pip install -r requirements.txt

📊 Dataset Pipeline Overview

The dataset pipeline consists of two main stages:

1. Data Extraction: Downloads and preprocesses raw data.
2. Sub-dataset Building: Constructs train and evaluation datasets from raw data.

---

🔍 Data Extraction

This stage includes:

1. downloader: Downloads raw trace files from the sources.
2. split: Splits the dataset into smaller chunks for efficient processing.
3. docking: Extracts unique docking stations.
4. raw: Extracts and stores trips in raw format (start station, end station, time, duration).

Each step can be executed independently using:

python main.py <command_name>

To execute all extraction steps for a given year:

python main.py all 2022

🌦️ Extracting Weather Data

To fetch weather data for a specific time range:

python main.py weather new_york 2022-03-01 2022-09-01 --collection-name observations

---

🏗️ Sub-dataset Building

This step constructs the final dataset from raw trips. The subdataset command is used for this purpose.

📌 Train Dataset Command

python main.py subdataset "none" 10,20,40 data/dataset citibike --name-suffix training --min-date 2022-03-01 --max-date 2022-08-01

🎯 Evaluation Dataset Command

python main.py subdataset "none" 10,20,40 data/dataset citibike --name-suffix evaluation --min-date 2022-08-01 --max-date 2022-09-01

---

🌍 Full Dataset with Zones

1. Generate zones:

python main.py zones 51 34 data/zones/ny --filter-region 71

2. Use zones to create the train dataset:

python main.py subdataset "39,0,28,41,31" 5-zones data/dataset-zones citibike --name-suffix training --min-date 2022-03-01 --max-date 2022-08-01 --nodes-from-zones --zones-path data/zones/ny/zones.json --add-weather-data --weather-db weather-new_york --weather-collection observations

3. Generate the evaluation dataset:

python main.py subdataset "39,0,28,41,31" 5-zones data/dataset-zones citibike --name-suffix evaluation --min-date 2022-08-01 --max-date 2022-09-01 --nodes-from-zones --zones-path data/zones/ny/zones.json --add-weather-data --weather-db weather-new_york --weather-collection observations

---

📌 CDRC Dataset

To access CDRC bike-sharing data, apply at CDRC for access to meddin-bike-sharing-world-map-data for the following cities:

• Dublin
• London
• Paris
• New York

📂 CDRC Dataset Structure

Each city dataset consists of the following files:

1. <city>_bikelocations.csv - Docking station details, including location changes over time.
2. <city>_ind_<year>.csv - 10-minute interval observations.
3. <city>_ind_hist.csv - Historical observations (similar to file 2).
4. <city>_sum.csv - Aggregated system statistics (every 10 minutes).

🔹 Identifiers: ucl_id and tfl_id are unique across files.

🏗️ CDRC Data Processing Pipeline

The CDRC processing pipeline consists of:

1. docking - Extracts unique docking stations.
2. raw - Extracts and stores raw trips.
3. zones - Creates geographical zones.
4. subdataset - Merges raw data, docking station info, and zones.

The entire pipeline can be executed with:

python main.py --cdrc all all <city> --dataset-path data/cdrc_data/datasets --nodes-from-zones --min-date <start_date> --max-date <end_date> --name-suffix <suffix>

🇮🇪 Dublin Dataset

Training:

python main.py --cdrc all all dublin --dataset-path data/cdrc_data/datasets --nodes-from-zones --min-date 2021-09-01 --max-date 2022-08-01 --name-suffix training

Evaluation:

python main.py --cdrc all all dublin --dataset-path data/cdrc_data/datasets --nodes-from-zones --min-date 2022-08-01 --max-date 2022-09-01 --name-suffix evaluation --skip docking,raw,zones

🇬🇧 London Dataset

Training:

python main.py --cdrc all all london --dataset-path data/cdrc_data/datasets --nodes-from-zones --min-date 2021-09-01 --max-date 2022-08-01 --name-suffix training

Evaluation:

python main.py --cdrc all all london --dataset-path data/cdrc_data/datasets --nodes-from-zones --min-date 2022-08-01 --max-date 2022-09-01 --name-suffix evaluation --skip docking,raw,zones

---

🎯 Summary

This repository streamlines bike-sharing dataset processing for research and analysis. It supports multiple cities, integrates weather data, and allows for structured dataset generation for training and evaluation.

For further details, run:

python main.py --help