Training and validation accuracy

I am working on project with independent Training and Test data set. I split my training data into training and validation data and accuracy score on my test data is higher than training data. I used Logistic Regression and DecisionTreeClassifier test accuracy is higher in both the cases. Shouldn’t test accuracy be lower than training accuracy ?

Hi. Could you let us know exactly by how much the test accuracy is better than the training accuracy?

This post from stack exchange should help you resolve your issue: