Kats, a kit to analyze time series data, a lightweight, easy-to-use, generalizable, and extendable framework to perform time series analysis, from understanding the key statistics and characteristics, detecting change points and anomalies, to forecasting future trends.
I suspect there is a bug in TSFeature, in trend_num_increasing feature. It seems to count the number of decreasing trends instead of the number of increasing trends.
This can be seen in the code, tsfeatures.py, lines 1631-1634:
for p in tdetected_time_points:
if p.trend_direction == "decreasing":
num_decreasing += 1
trend_detector_features["trend_num_increasing"] = num_decreasing`
Hi,
I suspect there is a bug in TSFeature, in trend_num_increasing feature. It seems to count the number of decreasing trends instead of the number of increasing trends. This can be seen in the code, tsfeatures.py, lines 1631-1634: