Data Set Collection for Flow Delegation

The data set collection is a zip file that contains 17 sqlite database files that can be inspected with any sqlite capable database reader (such as https://sqlitebrowser.org/). The folder names in the unzipped file indicate the names of the data sets (from d20 to d5050). Each database consists of a single table called "statistics" that gives access to the scenario parameters and evaluation results. Each row in the table represents a single execution of the evaluation environment (i.e., one experiment).

The columns starting with scenario are the parameters used for scenario / experiment generation. All other columns except for id and resultid (those two columns are not essential to the data set and can be ignored) refer to statistics gathered for one experiment. Columns starting with json contain a serialized json object and need to be de-serialized, e.g., by something like arr = json.loads(string) if python is used where string is the content from the column and arr is an array of floating point numbers. These columns contain time series data, i.e., the statistics were gathered for multiple time slots.

Example code for processing the data sets can be found at https://github.com/kit-tm/fdeval (plotters folder). The GitHub page also contains additional details about the data sets in this collection.