test target is input

[graeme44]

You seem to be using the test target values as inputs to the model - surely not a good idea.

[europa502]

Sorry for replying so late.

Yeah I understand the problem. I'm just a newbie still figuring out a way to rectify this issue.
I thought of crude way to fix this by varying the input value to match the output value from the trained model. So after some experiments I found out that if the test values(input values) are in the close range of output value, then they share some similar properties. So, maybe, if we provide a range, in which a certain input value coincides with the output value , we would be getting the approximate prediction.

Need to know if someone has any better idea.