How to initialize the class in MATALB?

1 次查看（过去 30 天）

显示更早的评论

Adan91h 2018-7-5

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/408910-how-to-initialize-the-class-in-matalb

编辑： per isakson 2018-7-8

在 MATLAB Online 中打开

I want to use 'fitrgp' function in Matlab Simulink.

fitrgp returns a GPmodel and its type of class is 'RegressionGP'.

for example:

y=fitrgp(rand(100,2),rand(100,1));
class(y)
ans =
    'RegressionGP'

i used the matlab function block in MATLAB simulink and used the following code

function y = fcn()
coder.extrinsic('fitrgp');
y=fitrgp(rand(100,2),rand(100,1));

it gives me following error and recommends me to preinitialize the output.

Function output 'y' cannot be an mxArray in this context. Consider
preinitializing the output variable with a known type

how can i preinitialize this 'RegressionGP' class so that i can use 'fitrgp' command in MATLAB Simulink?

2 个评论
显示无隐藏无

Wooshik Kim 2018-7-5

you can try initializing y

what is the output of fitrgp?

if it is an array, initialize y with an array of the same size

Adan91h 2018-7-6

在 MATLAB Online 中打开

Hi Kim,

No it is not an array. The output is a ''class'' with type 'RegressionGP'.

see the below response

>> y=fitrgp(rand(100,2),rand(100,1));
>> y
y = 
RegressionGP
           ResponseName: 'Y'
  CategoricalPredictors: []
      ResponseTransform: 'none'
        NumObservations: 100
         KernelFunction: 'SquaredExponential'
      KernelInformation: [1×1 struct]
          BasisFunction: 'Constant'
                   Beta: 0.5041
                  Sigma: 0.2552
      PredictorLocation: []
         PredictorScale: []
                  Alpha: [100×1 double]
       ActiveSetVectors: [100×2 double]
          PredictMethod: 'Exact'
          ActiveSetSize: 100
              FitMethod: 'Exact'
        ActiveSetMethod: 'Random'
      IsActiveSetVector: [100×1 logical]
          LogLikelihood: -11.8893
       ActiveSetHistory: []
         BCDInformation: []
Properties, Methods