O'Reilly logo
  • Jasper Brandes thinks this is interesting:

After we train this model, we can compute the cost according to some success merits on our training set. The resulting cost on this dataset is actually quite high, which is likely due to the fact that we did not properly preprocess and scale our input data, which we cover in depth in Chapter 25:


Cover of Spark: The Definitive Guide


success metrics?