I have a medical dataset with 14000 rows dataset with 900 attributes. I have to predict disease severity using that. I would like to know whether we can write rules in python language for training an agent for medical diagnostic using machine learning.

Can an agent make the decisions by the rules coded in python and that agent get trained with some machine learning algorithms? If so is there any agent architecture and model for the agent which is good in this context?

Edit: By the rule, I meant something like this.."if x>y output z as action". By the word "Training" I meant "how to tell this agent to do this action"?


You could formulate the problem of predicting disease severity as a classification one, you give the algorithm those 900 attributes and their corresponding labels (severe/not severe) after training, you give it a new data point with just the 900 attributes, it returns the correct label severe or not.

There is an enormous number of algorithms in the ML literature for classification, some of them formulate the problem explicitly as learnable rules, i.e. let the machine figure out the rules given the attributes.

  1. Classification and Regression Trees by Breiman et al (1984)
  2. Random forest by Ho in 1995
  3. XGBoost
  • $\begingroup$ "it returns the correct label": not always. $\endgroup$ – pasaba por aqui May 16 '18 at 6:59
  • 1
    $\begingroup$ yes, I mean in a probabilistic way $\endgroup$ – Fadi Bakoura May 16 '18 at 7:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.