Input and Output Data Construction

The City Of Knowledge

ai analytics datascience deep laerning machin python

description

In the last session, we discussed how to compute the training set size based on the batch size. Now, we're going to construct our input and output data for the LSTM. But first, we're going to do some pre-processing. First of all, we are increasing the upper boundary of our training set size. We are adding here timesteps multiplied by two. Why it's required, we're going to see in the next steps. So, now we just execute the cell. And we see our training set data goes up to 6,996, so we have increased by 20. The next step is very important for every neural network. We do feature scaling. So, this means we scale down every value which should be in the range between zero and one. To do this, we import a MinMaxScaler from sklearn processing package. We initialize this class MinMaxScaler and give the range from zero to one. And then we call the fit transform method from sklearn. This method takes our training set. And here we also cast to float 64 because otherwise, it will show warnings. So we do this. And now our data is scaled. The initial lectures series on this topic can find in the below links: Introduction to Anomaly Detection https://www.youtube.com/watch?v=IFHX4HUAo1w&list=PLpW3QouFxOnM6YWVOrcUBaQiy8EWi05pi&index=37

created

2023-09-24

staked

0.0 LBC

license

Copyrighted (contact publisher)

File size

21328645 Bytes