Kfold remains 0 after fitting

Question

Huyen Vu 2023-2-10

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1910315-kfold-remains-0-after-fitting

评论： Huyen Vu 2023-3-6

matlab.jpg

why does my kfold be set to 0 after beeing trained and set like that?

mdl = fitctree(T,"HH_kW_01",KFold=7);
mdl
%kfLoss = kfoldLoss(mdl)
display("end");

mdl =

ClassificationPartitionedModel

CrossValidatedModel: 'Discriminant'

PredictorNames: {1×62 cell}

ResponseName: 'HH_kW_01'

NumObservations: 35136

KFold: 0

Partition: [1×1 cvpartition]

ClassNames: [0 0.0049 0.0097 0.0292 0.0312 0.0321 0.0331 0.0351 0.0360 … ]

ScoreTransform: 'none'

Properties, Methods

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Piyush Patil 2023-3-3

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1910315-kfold-remains-0-after-fitting#answer_1184615

Hello Huyen,

It seems like the KFOLD property of your ClassificationPartitionedModel object is set to 0. Possible reasons for this to happen could be -

The KFOLD argument was not correctly passed to the "fitctree" function when the model was trained. So, to ensure that the KFOLD argument was passed correctly to the "fitctree" function, you can refer to the following documentation - Fitctree Function
The KFOLD property was modified after the model was trained or the model was not trained using cross-validation.

If the issue still persists, then please share the relevant code and information about data files. It will allow me to better understand the issue so that I can assist you in resolving it.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Huyen Vu 2023-3-6

Thanks for your answer. It's not only the problem with fitctree but also with other classifier such as fitcdiscr or fitcensemble. The issue always exists when I try to use cross validation, seems like the Kfold property was changed after beeing trained. So, about the data: 3500x64 table containing time series data about electric power consume.

请先登录，再进行评论。