When there is a prediction model, apart from R-square, residuals etc. what are the different ways of performing validation of the prediction model? Basically,what are the different ways (apart from R-square, residuals etc.) where you are validating of your Analytics prediction model given that the real/actual values will be available only in future ?
when you validate a model, you should first consider model use fit the intended business purpose or not. Is the method used appropriate for the question, is the data appropriate for the model, is the model parameter estimation consistent with the framework, is the model implemented correctly, is the ongoing monitoring results consistent with expectations. Each of these questions will have some tests you can perform. You need to be able to defend the model when others ask these questions.
Model evaluation and validation measures vary depending on the modelling technique you use.
The solution is cross-validation. See https://www.datasciencecentral.com/page/search?q=cross+validation