如何获得时间序列LSTM keras模型的“平均值”？

Question

So just an FYI, I have a pretty limited understanding of the mechanics of machine learning, LSTM, and time series modeling, but based on my current understanding, I feel like since I have a LSTM time series model trained on many time series plots, I should be able to get its "average" time series based on all of the ones it's trained on.

What's the best way to accomplish that?

I have a keras Sequential model, and I don't know if any code would even be helpful in this instance, but if there is any code that would assist, let me know!

EDIT: Here is some of the data

32.1576
31.92
31.7
31.85
32.05
32.5
32.3
31.975
31.7
32.15
32.6
32.55
32.4
32.4835
32.25
32.15
32.25
32.45
32.4
32.5002
32.45
32.5
32.5752
33.1748
33
33.35
33.45
33.45
33.425

Thanks!

Answer 1

You can use keras.layers.GlobalAveragePooling1D in order to average the time series. If the input shape is (batch_size, steps, features) , the output shape is (batch_size, features) . Weather or not you are using a Sequential model doesn't matter in this case. This layer works juts like any other layer in Keras. More information here: https://keras.io/layers/pooling/#globalaveragepooling1d