Why can't I use mapcaplot?

Hello,

I'm trying to preform PCA analysis for a set of values that I get from the attached excel.

I need to preform the analysis on all 40 variables.

clc
clear all
close all
x=readtable('PCA table.xlsx');
x=table2array(x);
mapcaplot(x)

When I try to use mapcaplot I get the following error:

Index in position 2 exceeds array bounds (must not exceed 1).

Error in mapcaplot>localUpdateBrushPlots (line 201)
ydata = data(:,INDEX_DIM2);

Error in mapcaplot>localInit (line 178)
localUpdateBrushPlots(fig);

Error in mapcaplot (line 65)
    localInit(fig,varargin{:});

Error in Untitled (line 7)
mapcaplot(x)

I would appreciate some help. I've never tried to do PCA before and I'm not sure what exactly the function needs as input.

Thank you,

Ruth

[Star Strider]

If you use the rmmissing funciton on ‘x’, only two rows of valid data remain.

Answer 1

[Adam Danz]

See the documentation.

The data input should be a numeric array or a DataMatrix object. Yours is a table.

Also see table2array.

[Ruth Pikovsky]

Hi, thanks for the answer. I did use table2array in my code.

[Adam Danz]

Sorry, I missed that. The renaming of x threw me off.

The NaN values in your data are the problem (as Star Strider indicated). mapcaplot calls pca() which removes observations with NaN values which results in losing entire rows of data that contain one or more NaN values.

I recommend cleaning up your data, specifically, filling the missing values, if possible.

[Ruth Pikovsky]

Thank you, the function works now. I'll try to figure out if I can replace the missing NaN values with zeroes.

[Adam Danz]

That will likely affect the pca results, though. Try interpolation if that makes sense for the data.

[Ruth Pikovsky]

Yeah that's also a good idea. Thank you.