saveCompactModel returning error with Compact Classification Tree

1 次查看（过去 30 天）

Tesla. 2017-5-1

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/338223-savecompactmodel-returning-error-with-compact-classification-tree

评论： Tesla. 2017-5-1

Hi, I am trying to save a classification tree model with saveCompactModel in order to generate C Code to predict new data as shown here: https://de.mathworks.com/help/stats/c-generate-code-and-classification-learner.html

My model is a Complex Decision Tree using the split criterion "deviance". Surrogate splits are set to "off" und advanced options. However, I still keep getting the error:

Surrogate splits are not supported for codegen. I am using R2016b.

Any ideas how to solve this problem?

Thanks!

2 个评论
显示无隐藏无

Gokhan Atinc 2017-5-1

Hi,

Decision Trees are actually not supported for code generation in R2016b. In R2016b, predict functions for only support vector machines (models trained via fitcsvm), linear models (models trained via fitclinear) and ECOC models using SVM's or linear models (models trained via fitcecoc) are supported for code generation. Decision trees are supported for code generation in R2017a, but trees with Surrogate Splits are still not supported for code generation in R2017a.

Gokhan

Tesla. 2017-5-1