0
$\begingroup$

I am currently doing a project in which the dataset is a lung cancer dataset. There is a training file which consists of 7 unnamed parameters (Attributes) and each of them have around 1000 values which are binary.

for example the file is somewhat like this,

0 1 0 1 1 0 1

1 0 0 1 1 0 0

0 0 0 1 1 1 1

1 1 0 0 0 0 1

This is just a sample representation. The original file has thousands of such values. There is a target file which consists of target values which can be either 1 or -1. 1 indicates that there is lung cancer and -1 indicates that lung cancer is not present. The target values are for each rows of the training set. I want to form a causal graph from this much data given which should use the Bayesian approach. Any other approach is also appreciated. Can someone please enlighten me on this problem.

$\endgroup$
1
$\begingroup$

You may start to look at the examples from BNLEARN here

Learn a BN model can be viewed as two steps

Step 1 is getting the structure / getting conditional independence for your random variables.

Step 2 is getting the parameters for a given structure.

Structure learning and Parameter learning two chapters clearly tell you what to do. In addition, the data set included in this R package is very similar to your data.

$\endgroup$

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.