Error when fitting polyfit curve

Trying to plot a polynomial curve onto my time series but keeps coming up with 'Inputs must be floats, namely single or double'

This is the code I've got:

% plot time series

X = Months;

X = categorical({'March','April','May','June','July','August','November'});

M = reordercats(X,{'March','April','May','June','July','August','November'});

Y = AverageDO;

plot(M,Y,'*');

p = polyfit(M,Y,1);

The excel sheet attached is the timeseries data.

Answer 1

[dpb]

Error is exactly what it says--you wrote:

X = categorical({'March','April','May','June','July','August','November'});
M = reordercats(X,{'March','April','May','June','July','August','November'});

so M is a categorical variable which polyfit is not equipped to handle.

Use

>> M=month(datetime(string(X),'inputformat','MMM'))
M =
     3     4     5     6     7     8    11
>> 

or some other way to generate the months as numeric.

I THINK some of the Statistics and/or Curve Fitting toolbox routines may know how to handle categorical variables but I don't recall for absolute sure otomh.