Hi Sierra,
I understand that you want to use the first column, that is, the ‘vertices’ column from the table, which is of the type of cell array, as the response variable for Gaussian Process Regression.
In this case, when you are using the “fitrgp” function, it is required that the response variable be a double vector or single vector. Gaussian process regression in MATLAB assumes a single response variable for each observation.
The column could be converted into a double vector and used as a response variable. To achieve this, you can use a method known as "aggregation" or "summary statistics". This entails using a selected measure or summary statistic to summarize the vector entries into a single value.
Here are a few common metrics you can consider:
- Mean
- Median
- Sum
- Maximum or Minimum
- Standard deviation
Here's an example of how you can convert the column into a double vector using the mean as the summary statistic:
c = cell2table({[1,2,4], 2, 4;[2, 3], 3, 5; [10, 20, 5,6], 15, 6});
c.Var1 = cellfun(@mean, c{:, 1});
gprMdl = fitrgp(c, 'Var1', 'KernelFunction','ardsquaredexponential',...
'FitMethod','sr','PredictMethod','fic','Standardize',1);
Please refer to the following documentation for more information on ‘responseVarName’ parameter of the “fitrgrgp” function:
You may also refer to the help documentation for the "fitrgp" function in MATLAB using the following command:
help fitrgp
Hope this helps to resolve the error.
