Bike sharing systems offer an affordable or even free means for users to navigate urban areas. Users can conveniently pick up a bicycle from one of the city’s distributed stations and return it elsewhere. While these systems have advanced, incorporating sensors to track user interac- tions, their management and data utilization often fall short, resulting in insufficient bicycle availability at stations. The growing demand for car rentals has prompted the modernization of bike sharing systems through automated processes. Worldwide, there are over 500 thou- sand bicycles in various bike-sharing programs, attracting significant interest for their potential to alleviate traffic, environmental, and health concerns. Currently, Spain, Italy, and China lead as bike-friendly countries with 132, 104, and 79 programs, respectively. The number of metropolitan areas adopting bike-friendly systems is on the rise [1].
This report focuses on bike sharing data, specifically collected for the years 2011 and 2012 in both hourly and daily scales. Our analysis centers on the daily time series, spanning 730 days, where we examine 10 out of 14 variables due to the correlation of bike-sharing rental processes with seasonal conditions like weather, precipitation, temperature, holidays, and days of the week. The dataset originates from the two-year historical records of the Capital Bike Share system in Washington D.C., USA.
The objective of this project is to utilize a linear regression model to analyze the dataset and predict the number of daily rentals within each time frame. Our investigation aims to unveil the relationship between the dependent variable (response variable) and environmental and seasonal variables acting as predictors. All relevant code is included in the attachment.