So as far as I understood, if we have a 0 for a phase in a process in recipe_metadata, that phase wasn’t applied, so all the rows corresponding to it don’t exist in the training set (for that process). If it is a 1 it was applied and it exists. However for 20005 in train_values, we have entries for “acid” and “final_rinse” while in recipe_metadata the code is 1 0 0 1 1, which means that we should also have a “pre_rinse” and all the rows associated with it, for the process 20005. Why are these rows missing? Is it a mistake in the recipe_metadata file or did I missunderstand the labeling? Thank you!
recipe_metadata is just a plan and does not always match the actual process.
In the announcement it says that:
Most of the time, the cleaning happens according to plan. However, for various reasons, in a small percentage of cases the cleaning that actually happens differs from the prescribed cleaning recipe.
Ohh, I see, thank you! But then, the recipe_metadata seems kinda useless. In the training set we can directly see the phases that actually took place (regardless of the plan) and for the test set, as we don’t know which cleanings took place according to the plan, the recipe_metadata is again not very useful. Am I missing something about the importance of recipe_metadata file?