Empty Features DF

[OmerBerkman]

I have a dataset of a few hundreds samples, each has about 7500 time points.
Feeding the algorithm with these 5M rows was too heavy so I took a toy example of 15 samples of each label (with most time points).
This results ~240K rows (2 * 15 * 8K).
I could extract a features DF of dimension 30*2349.

from tsfresh import extract_features
small_pd_extracted_features = extract_features(small_pd_timeseries, column_id="ix", column_sort="timestamp")

However, trying to "select" results in empty DF.

impute(small_pd_extracted_features)
features_filtered = select_features(small_pd_extracted_features, small_pd_y)

Does it mean the selection algorithm finds nothing correlated to the label, or have I done something wrong?

[OmerBerkman]

Update: I made it 40 samples (took more than an hour to calculate), and succeeded fetching 53 features. Exciting!