0

I'm doing multiple linear regression for a dataset.

The numeric_df dataframe is the continuous variables in the orignal dataframe.

I want to check linearity between the price variable (target) and the other continuous variables through pairplots. When I draw pairplots after imputing missing values, the data is like this:

enter image description here

And when draw pairplots before imputing missing values, the data is like this:

enter image description here

Is the relationship still linear after imputing missing values? Wil a linear model work on it?

4
  • Need more information. For example, what is the data like? How are you imputing the values? It appears that you're just imputing mean everywhere. Do you even need it, maybe getting rid of missing values would be better? Also, why do you assume the relationship is linear? Several relationships look exponential to me.
    – NotAName
    Commented Apr 24 at 6:49
  • To answer your question - I believe the linear model would work better without imputing missing values. There's too many of them and they will skew the relationship quite significantly
    – NotAName
    Commented Apr 24 at 6:51
  • 1
    Your question is not about a specific programming problem, a software algorithm, or software tools primarily used by programmers, hence off-topic for SO. A question about imputation effects (rather than its implementation) seems a better fit for Cross Validated. Please consult its how to ask section before posting, and it is bound to have similar questions there already. Also, please clarify what imputation method you are using. A minimal reproducible example is always appreciated.
    – ouroboros1
    Commented Apr 24 at 7:32
  • maybe you could ask on similar portals DataScience, CrossValidated or forum Kaggle - they should have better experience with Machine Learning
    – furas
    Commented Apr 24 at 9:56

0

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.