Hands-On Industrial Internet of Things

Defining the training set against the validation set

Normally, we split the data set, using 70% for training and 30% for testing:

X = df.valuessize = int(len(X) * 0.7)train, test = X[0:size], X[size:len(X)]

In this case, we cannot split the dataset to build a training model, because we do not know what is a normal operability and what is an anomaly. Therefore, instead, we decide to use an unsupervised method based on a moving average filter.