Validation and out-of-sample testing
These are all the available library functions that users have available for validation and out-of-sample testing. Refer to the corresponding usage page to see examples and explanations behind these functions.
EmpiricalRiskMinimization.lambda — Method.lambda(M) returns the lambda used during training. If multiple lambda values were specified, returns the optimal one
EmpiricalRiskMinimization.lambdaopt — Method.lambdaopt(M) returns the optimal lambda computed during training of M
EmpiricalRiskMinimization.lambdapath — Method.lambdapath(M) returns the vector of lambdas used
EmpiricalRiskMinimization.testloss — Method.testloss(M) returns the average testing loss
EmpiricalRiskMinimization.testlosspath — Method.testlosspath(M) returns the vector of test losses at various lambdas
EmpiricalRiskMinimization.thetaopt — Method.thetaopt(M) returns the optimal theta chosen during training
EmpiricalRiskMinimization.thetapath — Method.thetapath(M) returns the vector of optimal thetas at various lambdas
EmpiricalRiskMinimization.trainloss — Method.trainloss(M) returns the average training loss
EmpiricalRiskMinimization.trainlosspath — Method.trainlosspath(M) returns the vector of train losses at various lambdas