Predict response given new input data | GLME

Question

Joel 2018-10-11

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/423484-predict-response-given-new-input-data-glme

评论： Joel 2018-10-11

I am trying to fit a model between a dependent variable (Y) and two predictor variables (X, Z) using GLME.

First I divided the data into training (80%) and validation (20%). I've fitted a GLME model to the training data:

    X = Xtr % Validation portion of data (442x1);
    Z = Ztr % Validation portion of data (442x1);
    Y = Ytr % Validation portion of data (442x1);
    trTable = table(Y,X,Z);
    mdl_glme = fitglme(trTable,'Y ~ X + (1 | Z) + (-1 + X | Z)')'

Now I want to test this model on new (validation) data. This is where I run into problems. If I try:

    X = Xval % Validation portion of data (110x1);
    Z = Zval % Validation portion of data (110x1);
    Y = Yval % Validation portion of data (110x1);
    valTable = table(Y,X,Z);
    ypred = predict(mdl_glme,valTable)

I get the following error:

Error using categorical (line 434)
Unable to create default category names.  Specify category names using the CATEGORYNAMES
input argument.
Error in nominal (line 152)
            b = b@categorical(a,args{:},'Ordinal',false);
Error in classreg.regr.LinearLikeMixedModel/makeInteractionVar (line 806)
            G = nominal(ds.(interactionVars{1}));
Error in classreg.regr.LinearLikeMixedModel/extractGroupingInfo (line 205)
                model.makeInteractionVar(ds,interactionVars);
Error in GeneralizedLinearMixedModel/predict (line 539)
        ginfo = extractGroupingInfo(model,ds);

Though if I inspect trTable and valTable the only difference I can see is the number of rows:

    442×3 table (trTable)
      Y             X           Z 
      _________    _________    ________
       1.2107       1.2011      0.098847
   110×3 table (valTable)
      Y             X            Z 
      _________    _________    ______
       1.6064       1.6024      1.6024

I would be very grateful for any feedback and/or advice.

Regards J

2 个评论
显示无隐藏无

the cyclist 2018-10-11

Can you upload your dataset, or a minimal subset that exhibits the problem?

(Sometimes trying to find the minimal subset that causes the problem will actually help identify the cause.)

Joel 2018-10-11

Thanks! I really appreciate it! I will upload it first thing tomorrow morning as I do not have access to it at the moment (home computer).

Regards J

请先登录，再进行评论。

请先登录，再回答此问题。

Predict response given new input data | GLME

2 个评论
显示无隐藏无

回答（0 个）

另请参阅

类别

标签

产品

版本

Community Treasure Hunt

Predict response given new input data | GLME

2 个评论 显示 无隐藏 无

回答（0 个）

另请参阅

类别

标签

产品

版本

Community Treasure Hunt

2 个评论
显示无隐藏无