Oldenburger Fahrrad Challenge – Data Analysis

As part of the Oldenburg Bicycle Challenge, data will be collected from volunteer participants over a one-year period (September 2020 – August 2021). The project takes place within the European research project BITS (Bicycles and Intelligent Transports Systems), which is funded by the European Interreg North Sea Region (NSR) program. Partners from the leading cycling nations of the Netherlands, Denmark, and Belgium, as well as countries such as Germany and the United Kingdom that want to further increase bicycle use, are part of the project. The goals include increasing the share of cycling and reducing environmental impacts in the participating regions. The Oldenburg Bicycle Challenge pursues a gamification approach as an incentive for participation in the data collection via the smartphone app Ciclogreen. Here, prizes can be won by recording kilometers and users receive so-called Cycles points for kilometers cycled, which can be exchanged for gifts or discount codes at local merchants. The Oldenburg Bicycle Challenge as an overall project is divided into challenges of 1-2 months duration each.The first 6 challenges over the period from 16.09.2020 to 31.05.2021 were exploratively analyzed as part of a master’s thesis and the results are published on this website.

Analysis results

Characteristics of the bike challenges

ChallengePeriodTrip CountDistance (km)Duration (h)Ø Distance (km)Ø Duration (min)Ø Speed (km/h)Ø Age
1 – “Fit durch den Herbst”16.09.2020 – 15.11.20202,46617,9331,0487.2725:2916.7536
2 -“Schietwetter Challenge”16.11.2020 – 31.12.20202,77524,8761,5318.9633:0516.5246
3 – “Mit dem Rad ins neue Jahr starten”01.01.2021 – 14.02.20211,88917,2721,1309.1435:5415.7949
4 – “Outdoor-Challenge”15.02.2021 – 14.03.20211,36712,1347508.8832:5416.67
5 – “Leuchtend in den Frühling starten”15.03.2021 – 18.04.20211,54713,9958609.0533:2016.94
6 – “Klimaschutz im Alltag”19.04.2021 – 31.05.20211,83917,2841,0639.4034:4017.01
Total16.09.2020 – 31.05.202111,883103,4946,3808.7132:1316.6044
Tab. 1: Detailed statistics about each of the six challenges analyzed

Where do Oldenburgers cycle and with which speeds?

Fig. 1: Heatmap of the traffic volume recorded by the bicycle app

The heatmap shows the traffic volume within the central city area. The stronger the heat is mapped, the heavier the traffic volume was on the mapped roads. For example, roads colored dark orange have much higher traffic volumes than roads colored lime green.

The detailed maps can be explored for total data or seperated for each challenge by viewing them with a click on their name or download the maps as html file via the download button.

Fig 2: All recorded trips categorized by different speed levels

Slow rides are predominantly observed in the city, especially in the city center. The further out of the city the trips took place, the faster the potential cycling. However, the city center is not considered an exclusion criterion for fast cycling, since medium-fast and fast trips also often took place there.
The detailed maps can be explored for total data or seperated for each challenge by viewing them with a click on their name or download the maps as html file via the download button.

How old where the app users?

Information on the age of a user and a user ID could only be obtained up to the third challenge, after which the recording was discontinued.

Overall, the average age of users was 43-44 years. Users were thus on average 10 years older than in comparable studies evaluating smartphone-generated cycling data (Charlton et al. 2011; Lißner, Francke and Becker 2018).

The majority of users were between 26 and 35 years old (30.43%), followed by 46- to 55-year-olds (19.88%) and 36- to 45-year-olds (15.53%).

Users between the ages of 46 and 55 make by far the longest trips, averaging 13.54 km, followed by those over 66 years with 8.99 km.

Average speeds are highest among middle-aged participants (over 17 km/h). One young participant under the age of 18 drove slowly (13.2 km/h). In addition, it can be observed that speeds decrease with age. For example, 56- to 65-year-olds drive at an average of 15.6 km/h and older people over 66 drive much slower (13.8 km/h).

What was the gender distribution?

The number of users per gender could only be determined up to the third challenge, because subsequently the recording of a user ID was discontinued. Male users were slightly overrepresented with a share of 56.1%.

Male users recorded longer trips of 9.91 km on average, while the average trip length of female users was 7.73 km.

Since male users traveled further distances on average, they accounted for 63.26% of the total kilometers traveled.

When do the users cycle?

Users predominantly drive during the day. Significant peaks with over 1000 trips can be seen in the early morning and early evening at rush hour. Late in the evening as well as at night, hardly any bicycle trips could be observed.

Fig. 3: Heatmap: Amount of trips per weekday and hour of the start of the trip
Fig. 4: Heatmap: Amount of cycled kilometres per weekday and hour of the start of the trip

Weekends have only about half the number of trips per day as a weekday, and the fewest trips occur on Sundays.

Trips on Sundays are much longer than during the week, with an average length of 13.21 km. On average, 4.1 km longer trips are recorded on weekends than during the week.

On public holidays, the average trip length was 16.41 km, almost twice as high as on other days.

How many leisure trips are represented in the data?

Leisure trips were identified using a specially developed methodology. Thus, a route had to have at least one of the following characteristics to be considered a leisure trip:

  • Trip is outside of rush hour (Monday – Thursday: 04:00 – 09:00 and 15:00 – 17:00, Friday: 04:00 – 09:00 and 12:00 – 17:00)
  • Trip is on a weekend (Saturday or Sunday)
  • Trip is a round trip (samestarting), the length of which exceeds the average trip length of 8.7 km
  • At generell: Trip is at least 5 km long (except Sundays or holidays)

Using this methodology, 40% were identified as leisure trip, which represented 62% of the total miles traveled.

Leisure trips were much longer on average at 15.01 km than everyday trips at 5.34 km.

Leisure trips were not significantly faster.

What influence did the weather have on the cycling behavior?

In addition, the influence of the prevailing weather conditions at the start of the trips was investigated. Wind strength, temperature and precipitation were considered. Furthermore, the influence of snow was to be investigated, for which, however, no detailed data was available.


The wind forces were classified according to the Beaufort scale.

The average trip length does not decrease with increasing Beaufort strength, but only varies slightly regardless of the strength of the wind. No abnormalities were detected in the average travel speeds either.


For the purpose of investigating the influence of temperature on cycling behavior, temperatures were grouped into “<5 °C”, “5-15 °C”, and “15-25 °C”.

Users traveled similar distances, averaging 8 to 9 kilometers, regardless of temperature.


Precipitation intensities were categorized according to their intensities as described by the German Weather Service (DWD).

Both the average trip length and the average trip duration decrease in rainy weather. App users traveled 1.08 km shorter distances in rainy weather than in clear weather. Trip duration, which is strongly correlated with trip length, also decreases by 3:43 min from an average of 33:17 min to 29:34 min.

The heavier the precipitation, the lower the average trip lengths and trip durations of users. Rainy weather can thus have an impact on the length and duration of users’ trips.

Where and how often did users brake?

A time-sliding window approach with a time window of 5 s was chosen to identify the braking events, similar to the ECOSense project. Braking processes could only be identified for approx. 30 % of all trips, since a measurement frequency of 1 Hz is required for this purpose.

A total of 3-4 braking events occur per kilometer driven. The value varies in the course of the day. At night, from 0:00 a.m. to 8:00 a.m., there are considerably fewer braking operations than during the rest of the day from 9:00 a.m. onward. Even in the evening, the braking processes are still quite constant in the range of the average.

Fig. 5: Heatmap: Braking processes hotspots in the center of Oldenburg

The heatmap shows areas with a high amount of braking processes within the central city area. The stronger the heat is mapped, the more braking processes could be identified. For example, areas colored dark orange have much higher amount of braking processes than areas colored blue.

The detailed heat maps can be explored for total data or seperated for each challenge by viewing them with a click on their name or download the maps as html file via the download button.

How long did cyclists stand still during a trip?

Standing times could also only be estimated for recorded routes with a measurement frequency of 1 Hz.
The general assumption is that it is not immediately recognizable from the data for what reason the bicycle came to a standstill and whether a standstill was caused by a traffic light or for other reasons. Since GPS positions involve inaccuracies, it cannot be assumed that the speed is at exactly 0 km/h when the bike comes to a stop. A certain tolerance range should be taken into account, which, due to slight jumps in the position, also includes very low speeds for which a running drive seems unrealistic in the waiting times. This tolerance value was set to 4 km/h in the context of this work and thus corresponds to the lower limit of the average speed of a pedestrian.

The average standing time of 6 s per minute of travel time indicates a share of 10% of a bicycle trip.

The trend over the time of day is similar to that of the braking processes. At night from 0 o’clock to 8 o’clock in the morning, the standing times are generally shorter. In the course of the day between 7 a.m. and 6 p.m., they are in the high range of 6-8 s per minute of travel time. In the evening hours from 7 p.m. onward, there is a marked decrease to 4 s.

In which areas did slow speeds occur?

Fig. 6: Heatmap: Slow Speed problem area

The following heatmap shows the areas in Oldenburg, were the speeds are slowing more frequently and stronger. The more the heat is mapped, the more frequently and more heavily the speed fell below the average speed. The heat map can be explored only for total data by viewing it with a click on its name or download the maps as html file via the download button.


Charlton, Billy, Jeff Hood, Elizabeth Sall and Michael A. Schwartz. 2011. „Bicycle Route Choice Data Collection using GPS-Enabled Smartphones.“ Transportation Research Board 90th Annual Meeting.

Lißner, Sven, Angela Francke und Thilo Becker. 2018. „Big Data im Radverkehr: Ergebnisbericht: Mit Smartphones generierte Verhaltensdaten im Radverkehr.“ Technische Universität Dresden, Professur für Verkehrsökologie und Professur für Verkehrspsychologie