Dark Mode
Capital Bikeshare Demand Forecasting
Data Science and Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
The essential statistics needed for predictive modelling of urban bicycle rental demand. It captures the dynamics of a major bike-sharing system by detailing the counts of casual and registered rentals alongside key explanatory variables. The data facilitates research into mobility patterns and serves as a crucial resource for developing regression models to forecast future rental volume. Furthermore, its fine granularity allows for the detection and validation of unusual events or anomalies that impact daily travel.
Columns
- instant: A sequential record index.
- dteday: The specific date of the observation.
- season: Categorical classification of the season (1: springer, 2: summer, 3: fall, 4: winter).
- yr: The year (0 represents 2011, 1 represents 2012).
- mnth: The month of the year (1 through 12).
- hr: The hour of the day (0 through 23). Note: This field is only present in the hourly aggregation.
- holiday: Indicates whether the day is officially a holiday.
- weekday: The day of the week.
- workingday: Defined as 1 if the day is neither a weekend nor a holiday, otherwise 0.
- weathersit: Describes the current weather condition, ranging from 1 (Clear, Few clouds) to 4 (Heavy Rain/Snow + Fog).
- temp: Normalized temperature in Celsius.
- atemp: Normalized feeling temperature in Celsius.
- hum: Normalized humidity.
- windspeed: Normalized wind speed.
- casual: The count of casual users renting bikes.
- registered: The count of registered users renting bikes.
- cnt: The total count of rental bikes, comprising both casual and registered users.
Distribution
The data is available in two distinct files, typically provided in CSV format. The
day.csv file provides daily aggregated counts, consisting of 731 records. The hour.csv file provides hourly aggregated counts, featuring 17,379 records. The dataset structure includes time-based, categorical, and normalised weather features.Usage
This resource is ideal for several applications, particularly within the field of machine learning and data analysis:
- Regression: Predicting bike rental counts hourly or daily, factoring in environmental and seasonal settings.
- Anomaly Detection: Identifying and validating algorithms designed to detect unusual events in the city, such as those related to significant weather or public occurrences.
- Mobility Sensing: Utilising the data as a virtual sensor network to monitor mobility patterns and system usage within the urban environment.
Coverage
The data covers a two-year period, specifically 2011 and 2012. Geographically, it focuses entirely on the operations of the Capital Bikeshare system in Washington D.C., USA. The scope includes tracking the two distinct user groups: registered members and casual, unregistered riders.
License
CC BY-SA 4.0
Who Can Use It
- Machine Learning Developers: For training prediction models (e.g., Decision Trees) on time-series data related to transport demand.
- Urban Planners and Policy Makers: For assessing the impact of traffic, environmental conditions, and health initiatives.
- Data Analysts: For Exploratory Data Analysis and Feature Engineering exercises related to transport and weather dynamics.
- Researchers: Individuals interested in event detection and the complex interactions between human behaviour and meteorological factors.
Dataset Name Suggestions
- Capital Bikeshare Demand Forecasting
- Washington D.C. Bike Rental Data 2011-2012
- Urban Mobility Rental Count Data
- Hourly and Daily Bike Sharing Prediction
Attributes
Original Data Source: Capital Bikeshare Demand Forecasting
QUALITY