Meaning and usage of Residual Deviance in Analysis of Deviance Table after Logistic Regression



I was executing logistic regression in the Loan Prediction problem. After executing logistic regression, I came across an information that the a larger difference between null deviance and residual deviation is better for the model. However, I could not relate these terms with each other.

The code is as follows:

What inferences can we draw from this table regarding the importance of different variables? Please suggest some means to assess the accuracy of the model.


Lesser p-value indicates more significance i.e if some variable has P value 0.05 then we can say this variable has an effect on the model(dependent variable) with 95% level of significance.If the intercept is positive then it has a positive impact and if negative then it has a negative impact on the dependent variable.
For accuracy you can use ROC or confusion matrix or ks stats.
you can follow the following links as well:

for variable importance you can use this code:
varImp(model_name) # for u it may be varImp(logmodel)


Thank you for the answer. The p values will help us decide the significance of the variables. Could you also elaborate on the meaning of residual deviance with respect to this table and the insights from its values?

Thanks in advance :slight_smile:

hope this will help :slight_smile:


It is a great article. Thanks for the help.