This project provides a Docker images with a R environment and R scripts that implement the prediction algorithm (v2.0) for pilot SC4.
The structure of the input dataset should be like the following:
Link_id: The link_id of the OSM link
Direction: The direction of the link; (possible value is 1 or 2)
Date: The timestamp of the information
Min_speed: The minimum observed speed on this link/direction combo (in km/h)
Max_speed: The maximum observed speed on this link/direction combo (in km/h)
Mean_speed: The mean speed on this link/direction combo (in km/h)
Stdev_speed: The standard deviation of the observed speed values for this link/direction combo (in km/h)
Skewness_speed: Skewness of the observed speed values for this link/direction combo (in km/h)
Kurtosis_speed: Kurtosis of the observed speed values for this link/direction combo (in km/h)
Entries: The total gps signals that matched on this link/direction combo
UniqueEntries: The unique taxis that matched on this link/direction combo
The R module (which is installed as a package) takes this dataset as input, it trains a model for a link/direction and produces a prediction for the requested variable (i.e. Mean_speed).
The GetPrediction function (in the GetPredictions.R file) acts as a wrapper by calculating predictions for many links at once and for the next 4 quarters.
$ docker run --name forecasts -p 6311:6311 -d ptzenos/pilot-sc4-rscripts:v2.0
To run an example, you can do the following:
R < GetPredictions.R --no-saveA csv file containing the output should then be created in the folder "output".
TBD