Using weather data to understand usage of Glasgow’s cycle hire scheme
In late 2016, Karen Macpherson of the Glasgow Centre for Population Health approached UBDC for help with one of a series of reports on active, sustainable travel that the GCPH was preparing on Glasgow City Council’s cycle hire scheme. They wanted to analyse usage of the scheme against a range of factors, one of which was weather. While the Met Office produces the data they wanted, they requested our help in accessing and preparing it for useful linkage, which our expert team of data scientists was happy to do. The final report, Glasgow’s public cycle hire scheme: analysis of usage between 2014 & 2016, was published in March 2017. This case study provides a high-level summary; please check out the full report for a full analysis including graphs, charts, and factors other than weather!
Only 23% of households in Glasgow have access to a bike for personal use. Glasgow City Council’s vision is “to create a vibrant Cycling City where cycling is accessible, safe and attractive to all” (PDF 6.23MB). Supporting growth of the Council’s cycle hire scheme will help to make cycling a more viable option for many people. The cycle hire system in Glasgow was launched in 2014 and within the first two years of operation 16,000 people had registered for the service resulting in around 200,000 hires.
We wanted to explore how successful the cycle hire scheme has been in Glasgow, and what factors influenced public take-up - documenting its growth, peak times of use, demographic aspects, and seasonal and other external factors that had an impact on its use. We wanted to measure the impact of weather conditions as part of our analysis.
The Urban Big Data Centre supplied Met Office weather data from their Open Data Collection for us to use. They also helped us by adding value to the data by curating it first to make it easier for us to use for our purpose.
Weather data recorded by the Met Office’s Glasgow/Bishopton weather station provided hourly records indicating the observed temperature, wind speed, wind gust speed, and prevailing weather conditions. We linked the bike hire data to this data, enabling the individual impact of the different weather variables on number of hires made to be studied.
The overall mean number of bike hires over the two-year period was around 260 hires per day. We found that sunny weather does have an impact on the number of hires, with the highest average number of rentals occurring then. Numbers were around average in cloudy conditions and when showery, suggesting that these conditions don’t impact on people’s decisions to rent a bike.
Rain, on the other hand, does seem to reduce numbers, as does snow and mist/fog. Numbers for less prevalent weather conditions such as snow showers are small so the results shown for these instances are less certain than those for more commonly observed conditions. Likewise, the number of hires corresponding to weather observations during hours of darkness are smaller than those for daylight weather observations and again the results should be treated with caution.
Our findings on the impact of the air temperature on number of hires shows a strong relationship between temperature and frequency of hires. As the temperature increases so does the number of hires, with an increase in average number of hires per hour of 1.44 for every 1 degree increase in temperature. Again, the numbers at either end of the graph reflect small numbers of occurrences, so should be treated with caution. The number of hires per hour decreases gradually as the wind speed increases.
The data prepared for this project has been made available as a separate slice in UBDC’s Open Data Collection, to enable others to use it. This mutually beneficial collaboration is something the UBDC hopes we can repeat in future with the GCPH, and in fact we have another case study coming soon from Bruce Whyte on the use of our urban indicators data to investigate children’s access to greenspace in Glasgow – watch this space!