Plot a graph that interpolates between NaNs?

I'm plotting a line graph with two y vectors against an x vector (using plotyy). Both of the y vectors have data gaps as NaNs. When I plot I do not want there to be gaps on the graphs where there are NaNs, I want it to connect all the 'real values'.

A solution offered for others is to interpolate the values before plotting. This DOES NOT work for me because I want to have a marker point at each of the real values. I attach a picture of what I want created in Excel.

------ For simplicity I've described the problem as above and not provided code. In fact I'm plotting against 3 y-axis but the 3rd y-vector does not have NaNs. I'm using the user-created function plotyyy.m which itself calls MATLAB's function plotyy.

Thank you

[KSSV]

Using interp1 you can fill your gaps/ NaN's.

[Grant]

This doesn't work

See above: "A solution offered for others is to interpolate the values before plotting. This DOES NOT work for me because I want to have a marker point at each of the real values. "

Answer 1

[Stephen23]

So remove the NaNs:

id1 = ~isnan(y1);
id2 = ~isnan(y2);
plotyy(x1(id1),y1(id1),x2(id2),y2(id2))

[Grant]

But then the data does not correspond to the other y data or the x data. Unless I've misunderstood.

[Guillaume]

I don't understand what you mean by "the data does not correspond to the other y data or the x data".

Stephen's answer only plot those points that are not NaN, at the exact same location as your original data. This sounds exactly like what you asked.

[Stephen23]

@Grant: My answer does not change your data: the x values are the same, and the corresponding y values are the same. All my answer does is remove the NaNs. The plotted position of any (x,y) pair will not change. Try it.

[Grant]

Thank you very much.

(Apologies, I misunderstood and MATLAB was not available to me at the time)

Answer 2

[Steven Lord]

If you're using release R2016b or later, use the fillmissing function to fill the NaNs (which represent missing data for double and single arrays) using a method of your choice. Call fillmissing with two outputs to both identify the locations of the NaNs and fill them, then use the second output from that call to generate a vector to use for the markerIndices property of your plot. This will allow you to plot markers only for the points that were in your original data set, not the points that were filled.