Hi there,
Super interesting stuff! I reproduced your code and have a question about the results. After generating a fake Pima table, I compared its Phik correlation scores to that of the original and got something strange. The intra-feature characteristics of the fake table mostly match up with the real one but then for some reason, DiabetesPedigreeFunction has 1's across the board. Does this seem possible or have I perhaps made an error in following your code? If these Phik measures are correct, could that be why there was an increase in accuracy?
n
Hi there,
Super interesting stuff! I reproduced your code and have a question about the results. After generating a fake Pima table, I compared its Phik correlation scores to that of the original and got something strange. The intra-feature characteristics of the fake table mostly match up with the real one but then for some reason, DiabetesPedigreeFunction has 1's across the board. Does this seem possible or have I perhaps made an error in following your code? If these Phik measures are correct, could that be why there was an increase in accuracy?
n