Validation Techniques fro Logistic Regression



Can anyone briefly explain what are different Validation Techniques used for validating a model built by using Logistic Regression


Hi @ksmvsn

The easiest way without going by the likelihood or odds is to build the confusion matrix on you prediction and then to calculate accuracy, specificity etc …

Hope this help.



Hi @ksmvsn , there are various metrics to validate a model built by using Log Regression (which is basically a classification technique) like confusion matrix, concordant-discordant pair, Lift Chart, KS statistic, ROC curve etc.

Lift Chart, KS statistic tells us how our model is able to segregate between positives and negatives. You can read about the other metrics mentioned in the below link:

Hope this helps!!


The following measures of fit are available,

1.Chi-square goodness of t tests and deviance
2.Hosmer-Lemeshow tests
3.Classication tables
4.ROC curves
5.Model validation via an outside data set or by splitting a data set

link to the pdf file,

Hope this helps,


Hi @ksmvsn

there are multiple techniques and which particular techniques to be utilized in which scenario is important. Best guide to understand this is through I have mentioned the link below.
validation techniques like lift is mostly used when we are working on marketing campaign data. so best way to understand all is participate in Data Hack organized by AnalyticsVidhya team