分割验证线性回归connection error

1630717bremmers

So I have this project I have been working on and I splited my dataset with split validation to use linear regression on it, but to the connection etc I used the permormance regression after I used apply model, but the outcome of the proces is not quite right, I believe it's because the mod port from the apply model is not connected to the end, is there a way to fix that?

Thanks in advance.

lionelderkrikor

Hi@1630717bremmers,

"..but the outcome of the proces is not quite right..."

Could you be more precise and explain what you mean ?

Thanks you,

Regards,

Lionel

1630717bremmers

Hi Lionel,
The only outcome I have is the root mean squared error from the peformans operator. I'm not getting the prediction from the linear regression

lionelderkrikor

Hi@1630717bremmers,

Many things :

1. To visualize your model connect the output portmodofSplit Validationoperator to theresport.

2. To calculate and display other performance metrics (not only the RMSE), check them in the parameters ofPerformance(Regression)operator.

3. To calculate and display the prediction there are 2 ways :

a. Use the coupleRemember / Recallto recover the labelled dataset inside theSplit Validationoperator, like in this process (to adapt to your own data) :

<连接from_op = "应用模式”from_port = "标签data" to_op="Performance" to_port="labelled data"/>

b. Use aCross Validationoperator instead theSplit Validationoperator and connect the test output ofCross Validationtes

to theresport.

In deed, the performance calculated in a cross validation is considered as more representative of the real performance of your model on future unseen data.

I hope it helps,

Regards,

Lionel

Telcontar120

You can also just use Store to save the model, and then use Apply Model with it and any other dataset (including the original development dataset) in the future to generate the set of predictions using that model.