fitrm with NaN values

Hello: I am having difficulties fitting a repeated measures model (fitrm) for matrices that contain a large amount of NaN values. I have seen few posts in this regard but haven't found a way to resolve it. Any suggestions?

[KSSV]

What difficulty? You getting any error?

Answer 1

[Leila]

Yes, I get an error as below:

Subscript indices must either be real positive integers or logicals.

Error in classreg.regr.modelutils.designmatrix>dummyVars (line 460) X0(end,:) = -1;

Error in classreg.regr.modelutils.designmatrix (line 290) [Xj,dummynames] = dummyVars(dummyCoding{j},Xj,catLevels{j});

Error in RepeatedMeasuresModel.fit (line 1326) = classreg.regr.modelutils.designmatrix(ds(~missing,:),...

Error in fitrm (line 67) s = RepeatedMeasuresModel.fit(ds,model,varargin{:});

Error in multistats (line 62) rm = fitrm(NewT,strcat('stimnod1-stimnod',num2str(s),'~Var1'),'WithinDesign',STIMNOD)

Error in script_NHP_v15 (line 534) [tblAmp3,tbmAmp3,NewTAmp3,LabelStatsAmp3]=multistats(AmpStimOut2Nod,AmpStimOut2Neighb,AmpStimOut2Out);

[Kameron Clayton]

Was a solution for this issue ever found?

[Kameron Clayton]

            % Rows with missing values cannot be used in the fit
            missing = findMissing(ds,formula.Terms,responseCols);

It seems the answer is in fit_rm code ln. 1307