I have an independent variable with 19 dimensions(19 features) and I need to perform stepwise regression. I need to perform iterative regression because the target value I am predicting becomes available in a timely manner and I need to make a prediction at each iteration with available independent variable values and dependent values. When a new dependent value becomes available, I need to use that value to update my model. This is, in essence, a special case of Kalman Filter. I could use statsmodels.tsa.kalmanf but it is more convenient for me to look at it as a regression problem. I am currently using an implementation of the regression update approach stated in https://www.jstor.org/stable/24305577?seq=1#metadata_info_tab_contents for multi dimensional independant variable.

  • $\begingroup$ You may want to have a look at sklearn implementation of passive aggressive regressor: scikit-learn.org/stable/modules/generated/… $\endgroup$ – kanimbla Apr 9 at 21:45
  • $\begingroup$ How many observations do you have? With only 19 features, it may well be feasible to just rerun the whole stepwise regression as you get new data. $\endgroup$ – jbowman Apr 10 at 0:36

Your Answer

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

Browse other questions tagged or ask your own question.