Bad @brief in energy_ratio_by_chunks

[Nexxel]

Describe the bug
The actual @brief of energy_ratio_by_chunks in khiva_c is
@brief Calculates the sum of squares of chunk i out of N chunks expressed as a ratio.

• with the sum of squares over the whole series. segmentFocus should be lower
• than the number of segments

Expected behavior
The @brief should be:
@brief Calculates the sum of squares of chunk i out of N chunks expressed as a ratio

• with the sum of squares over the whole series. segment_focus should be lower
• than the number of segments

[BilguunBatsaikhan]

@Nexxel excuse me, but do you know how is segment_focus used in the calculation?

[avilchess]

Hi @billydentsu,

This operation divides the sum of square elements within a chunk by the sum of square elements of the whole time series. Thus, the segment_focus parameter receives the segment number you want to compute.

Cheers.

[BilguunBatsaikhan]

Hi @avilchess, I see. Thanks for explaining.
I have a trained a model, and it is most significant feature seems to be energy_ratio_by_chunks where the segment_focus is the last chunk.
If I may, does this mean that the last few series (values of the last chunk) of my time-series are the most predictive?
Does this make sense?

[avilchess]

@billydentsu
I´m not sure what model are you using, so I may need a better description of your scenario (data, purpose, and models) to provide an accurate answer. I would say that the last chunk of your time series is the one with more conserved energy (w.r.t. the whole time series). Having said that, If your model is paying special attention to this parameter, probably It is having an impact in the classification or segmentation of your training dataset.