I'm doing machine learning on machine data and trying to predict whether my machine will be knocked out or not.
I have difficulties to use the Machine Learning Toolkit module since I try to predict rare events (knockouts). As a result my algorithms predict almost always OK and the accuracy is good, while for me it is very bad.
(1) I have tried to add new templates in the module configuration files but I can't do it. I don't know where to put my commands from scikit learn in the .py script (in init or fit, I don't really know the difference).
(2) I also wanted to add a preprocessing to allow data resample but it's the same as problem number 1.
(3) I would like to review the way the module cuts the train and the test and I don't know if it's possible. I don't want it to be random, I want my classes to be balanced in the train.
(4) I would like to do cross validation but it's the same problem I don't know if it's possible.
I have a lot of questions so if you could help me I would like to do it !
Have a nice day !
Have you also installed the Python for Scientific Learning App as well? You can add your custom algorithms there.
If you want to make any changes in teh algorithm, you can copy it and make the changes in the new file.
Hey, thank you for your answer. Yes I needed the Python add on to install the machine learning toolkit. I tried to make the changes in the algorithms but when I look at the scripts, I never know where to do my modifications. That's why I posted this question. For example, I wanna add "weight" to the KO versus the OK.