How to calculate MAPE, MAE and SD for LRN neural networl

Dear all,

I am currently using LRN neural network for the prediction and needs to calculate performance indices like MAPE, MAE and SD. As mentioned below, i am trying to calculate from the given formula but i am getting the value in double which is undesirable as the value need to be in an integer form although i am getting the value of other performnance indices like MSE and RMSE in integrer form. Please assist in this regard

lrn_net = layrecnet(1,10);

lrn_net.trainFcn = 'trainlm';

lrn_net.trainParam.show = 5;

lrn_net.trainParam.epochs = 500;

lrn_net = train(lrn_net,data_train,target_train);

y = lrn_net(data_test);

performance = perform(lrn_net,target_test, y)

RMSE = sqrt(performance)

MSE = mse(lrn_net,target_test, y)

MAPE=(sum(abs((y-target_test)./(y)))./length(target_test))*100

MAE=(sum(abs(y-target_test))./length(target_test))*100

SD=(std(y-target_test))*100

Answer 1

[Kyle Rawding]

Hi Shaheer -

To convert from double to int, you could use the uint8 or int8 functions, or another from that family of functions.

For example,

MAPE=uint8((sum(abs((y-target_test)./(y)))./length(target_test))*100);

Hope that helps,

Kyle

[Shaheer Ansari]

Thank you for your assistance in this regard. It'll be helpful to me in your research work