Logistic Regression Variable Importance/Factor Strength

Hi everyone,

After building a model in Knime, is there a node or way I can visualise and assess the importance and strength of each individual variable? 

I want to be able to justify why each variable has been included.

Hi jbarton,

are you specifically referring to the Logistic Regression model?

If you are interested in knowing which variable is significant among the ones that you have used, you should take a look at the output of the node, when the node has been executed. In other words, after fitting the model, you may want to examine the contribution of individual predictors. In logistic regression, the regression coefficients represent the change in the logit for each unit change in the predictor. So you may want to focus on a predictor's effect on the exponential function of the regression coefficient – the odds ratio (more info here: https://en.wikipedia.org/wiki/Odds_ratio) for easier interpretation.

If you are referring to which variables to include before to perform the logistic regression, then you would need to use some techniques that help you to use perform feature selection. For more details about this, I would point you to the following link: https://www.knime.org/blog/seven-techniques-for-data-dimensionality-reduction.

Hope this helps,

Best,

Vincenzo

Both useful, thanks Vincenzo