CODEWITHSUNDEEP
python-3.xmachine-learning
Baptiste Nguyen
Baptiste Nguyen

Reputation: 57

Machine Learning with some rules design by human

For my issue, I have the following machine: machine

In this machine, there is 2 doors where one product can enter. Here are some rules and observations:

-rule 1: If one product enter by a door, it comes out by the same door.

-rule 2: For a door, if a product has already entered the machine by it and hasn't yet come out, no other product can enter by it.

-Observation 1: If 2 products process at the same time on the machine, they process slower.

-Observation 2: The process time of a product depends on its type.

For the project, I need to create a model which estimate the process time of the different products in this machine. So I have decided to use machine learning with the following input and output: graph for machine learning

The issue is that: some time there is only one product in my machine. In my training dataset, if there is no product in one door (for exemple,door 1), the input and output are like this: P1=0;T1=-1;B1=-1;E1=-1

Does the Machine-Learning Algorithm of scikit-learn will learn this specificity? Or do I need to create 2 model: one where there is only one product in my machine and one where there is 2 product in my machine? Or is it possible to include some rules in the model (for example, if P1=0 then (E1=-1,don't do the training for the part of the model concerning door1 and do the training for the part of the model concerning door 2))?

Upvotes: 2

Views: 52

Answers (1)

ichafai
ichafai

Reputation: 341

If you have enough training examples, (and you train your model well) yes it will understand that when P1=0 and T1=-1 then it should predict something close to E1=-1.

Upvotes: 1

Related Questions