To properly analyze pedestrian and bicycle traffic, it’s important to keep a critical eye on the data collected, and to identify and analyze the causes that may explain positive or negative trends.
By way of introduction, here are a few reminders about data collection.
How does a bicycle/pedestrian counter work?
For over 20 years, we have been supplying systems for collecting reliable field data by recording the passage of users on site (urban or natural). This data collection is carried out using several technologies, capable of meeting all site constraints:
- induction loop technologies for cyclists
- pyroelectric sensors for pedestrians
- cameras with algorithms and artificial intelligence for multi-user counting.
NB: In the terminology of this article, we distinguish between the notion of “counter” for the physical equipment and “site”/”counting site” for the geographical location where the physical equipment is positioned (it being understood that it is possible to move counters from one location to another, and therefore from one “site” to another). The traffic measured is that of a “site”.
Analyze pedestrian and bicycle traffic
Once collected, the data can be analyzed. However, to be relevant when analyzing pedestrian/bicycle traffic, it is important to be aware of the various possible causes of changes in traffic levels, as well as the points to watch out for when making comparisons.
Here are a few resources on these causes and points to watch out for, identified with Stéphanie Mangin, Head of the Observation Department for Vélo & Territoires, Eco-Counter’s partner since 2013 in the creation of the “Plateforme nationale des fréquentations” (PNF).
Causes of changes in visitor numbers
The effects of the weather on cycling
While automobile traffic is generally very predictable, cycling and walking are more sensitive to the weather. Winter periods are traditionally weaker in terms of traffic, and meteorological events can have an impact on ridership.
A few remarks on weather effects (in France, according to observations made for the PNF):
- Cycling in urban areas is increasing year on year, and seems less and less sensitive to the weather.
- On the other hand, leisure/tourist cycling is stabilizing on a plateau at the high level reached at the end of the health crisis. The slightest change in the weather, whether positive or negative, from one year to the next, has an impact on visitor numbers.
To analyze attendance correctly, it is therefore important to have constant-weather data, or at the very least to isolate the impact of weather on the increase in practice. For this, cross-referencing attendance data with weather data can also help (cf. the weather analysis module integrated into our Eco-Visio solution).
Public holidays can also have a significant impact on practice. In urban areas, for example, a public holiday in the middle of the week can lead to a drop, whereas in rural areas, a public holiday in the middle of the week can lead to an increase. This potential impact on ridership is a factor to be taken into account when making comparisons with previous years.
Closure due to roadworks or changes to the traffic plan
Another more easily identifiable major cause is the closure of an arterial road for roadworks or transit traffic, which almost systematically leads to a change in ridership. This is an element to be taken into account when comparing ridership figures on a regional scale: is the analysis carried out on comparable sites between the two years analyzed?
The “cycling network” effect
On a more positive note, the creation of a cycle network can also lead to misleading traffic reductions. The development of a new, more suitable or more direct route can lead to a reduction in bicycle traffic on another route. In order to objectify this perception in cities where work is underway to build a complete and continuous cycle network, it is necessary to cross-reference the traffic data from permanent counters with other data sources, such as GPS tracks, to obtain a complete view of the network. A possibility offered by our Cycling.Insights solution, developed in partnership with Geovelo.
Key aspects to bear in mind when making comparisons
Comparisons with comparable periods and data
Before comparing a site’s data between two periods, it’s important to ensure that the data set is complete and does not present any anomalies over the two periods to be analyzed. If this is not the case, to avoid distorting the calculation of trends, it is important to carry out analyses for comparable periods. For example, if you wish to compare the years N and N-1 for a counter that is missing ten days in January of year N, the comparison will be made by excluding these ten missing days for both years, N and N-1. In this way, only the common period between the two years is taken into account for the evolution.
The need to compare what is comparable implies working on the most detailed data set possible. Only a fine-scale analysis (ideally by hour and by flow) can detect anomalies or missing data. It’s easy to see why: a counter that has supplied data for all 365 days of the year cannot be compared with one that has had missing days of data. This will necessarily result in a comparison that is meaningless, but which may be invisible if we focus on the total number of passages recorded for a year.
Similarly, a dip in the data may follow an abnormal rise, and compensate for it, giving a false impression of stability in the practice, when a closer look would show a different story.
Nature of cycling facilities
When analyzing ridership, it’s important to know the nature of the cycling facility: is it mono-directional or bi-directional? Does the facility have one or more lanes (e.g. a shared lane, and a dedicated lane separated from traffic)? Care should be taken when comparing “absolute” traffic figures between sites, to avoid comparing sites with different layout configurations.
Number of passages, rate of progression or evolution on a base of 100?
Aggregating the numbers of cyclists passing through several sites in order to communicate a “big number” for traffic over a year is not relevant, as it says nothing about actual practice, but only about the number of counters installed, multiplied by the average traffic. It is more honest and meaningful to communicate the average number of passages at a site, the evolution of this number of passages from one year to the next (integrating all the causes/limitations mentioned above), or, as we do for our analysis of bicycle traffic in 15 countries, a base 100 representing, for example, the average traffic for one year, which enables us to compare locations with different traffic levels, and to follow this evolution over time.
Finally, it’s worth bearing in mind that a 10% increase in traffic doesn’t have the same meaning depending on whether we’re talking about a counting site that registers 100 passages/day or 1,200 passages/day. It may therefore be relevant to mention, in addition to the % increase, the average traffic on the site.
About our data services
At Eco-Counter, we provide data expertise services tailored to your needs. Whether you need specific reports, validation, data reconstruction, or an estimate of annual or seasonal volumes from short-term counts, we can help.
Beyond data collection via our physical products, we offer a wide range of services that can help you better understand your data.
Here’s how we can help:
- Our data validation and reconstruction services ensure the accuracy and reliability of your data, so you can rely on the results. Using a proven and reliable methodology, we validate the consistency and integrity of datasets, reconstruct data where necessary, and communicate the results to you transparently (including our level of confidence in the results obtained).
- We can also create specific reports to provide detailed analysis of your data, giving you the information you need to make the right decisions, including weather data.
- In addition, we can also help you estimate annual or seasonal volumes by extrapolation from short-term counts. This is a valuable method for gaining a better understanding of your network with fewer resources. This unique method is based on meters with similar profiles in the database, allowing for the integration of local components and terrain peculiarities (weather effect, for example).