The datatest feature loads a dataset into ClickHouse, that is then used for calculating the values for the confusion matrix. However, at this point users can only use a dataset file called dgta_dataset.json.gz with very specific columns. We should allow users to insert (more) arbitrary dataset files, that have to contain only a limited number of specific columns. This should also work for several file types, like .json, .json.gz and .csv. Maybe we could also add the option to load data from multiple files at the same time.
The datatest feature loads a dataset into ClickHouse, that is then used for calculating the values for the confusion matrix. However, at this point users can only use a dataset file called
dgta_dataset.json.gzwith very specific columns. We should allow users to insert (more) arbitrary dataset files, that have to contain only a limited number of specific columns. This should also work for several file types, like.json,.json.gzand.csv. Maybe we could also add the option to load data from multiple files at the same time.