Which step we follow first in ml?

When we are dealing with ml problem. which step we follow first imputing missing value or analysis of feature (like plotting relationship graphs between input and target variables )

Hey @Abbas_Malik,
I think a third thing to do that you are missing here, which you could have asked is, plotting the distribution of the input features. In my opinion, if you start with this, it will help you to decide how to impute the missing values. For instance, if you find that the features follow a gaussian distribution, you can try to do some imputation based on this insight.

Now, let’s assume that we know how we are going to perform the imputation. Here, note that analysis of features may be done before imputing the missing values (by neglecting the missing values) and may be done after imputing the missing values as well.

For instance, if you do this analysis before performing any kind of imputation, you may get to know that certain features are not at all useful for predicting the target, hence, there’s no point in using them as input features, so, we can simply drop these.

Now, let’s say that you have performed the imputation on the remaining features. You can again perform this analysis (using features on which you have performed imputation), to know if they are still relevant or not, and if not, then once again, we can simply drop these.

Let me know if this helps.

Cheers,
Elemento