Non-aggregated time feature extraction

[ytzfhqs]

I now have the following data:
date v1 v2 Target
2018/1/1 0:00 2.2 1.3 244033.7682
2018/1/1 0:15 2.4 1.5 232720.6402
2018/1/1 0:30 2.7 1.7 231685.6891
2018/1/1 0:45 2.5 1.9 227581.3317
2018/1/1 1:00 2.6 2.3 233482.4269
2018/1/1 1:15 2.2 1.4 224464.7554
2018/1/1 1:30 2.6 1.7 223881.3222
2018/1/1 1:45 2.7 1.6 221522.7156
date is the time column, v1, v2 are the feature columns, and Target is the column to be predicted。

I would like to know if tsfresh can help me automatically build features to better predict the Target column, which is a regression problem

[nils-braun]

Yes, you can use tsfresh to build features for this use-case. We even have an example for this: https://github.com/blue-yonder/tsfresh/blob/main/notebooks/05%20Timeseries%20Forecasting.ipynb