Bike counts of Nantes Métropole – Years 2014 to 2019

History of bike counts: raw data, anomalies detected and adjusted values

This game provides, for 20 counting loops whose data collected between 2014 and 2019 are usable, the number of passages recorded each day. This corresponds to three variables: date, meter and data recorded. These data are supplemented by two additional variables: an anomaly indicator that reports whether the counting data for the day in question is likely to be incorrect and an adjusted counting value that corresponds to the estimate of the most likely number of crossings for the day in question.

The detection of anomalies and the estimation of an adjusted value have been established using an algorithm whose operation is also public. For the sake of transparency, two supporting documents are published in conjunction with the dataset:

- an explanatory note that explains the logic of the algorithm and describes its operation. This note includes an application of this algorithm on counting data opened by Nantes, but also those opened by Rennes and Lille.

- a detailed report that applies the algorithm to each of the counting loops and returns the results in the form of graphs and raw data.

Each of these two documents is an automatic report produced with free software (R). We also publish the source code of these reports so that everyone can analyse how the anomaly indicators and adjusted counting values have been calculated and reproduce the same results on their own computer to ensure their reliability.

The current year’s bike counting data are also available on this portal, as well as [the location of the counting loops].