Error in Matlab trying to perform mesh grid

Hi! I am relatively new to Matlab and have been trying to get an indepth look/analysis into my excel data set by performing contour plots and heat maps. I'm trying to extract individual columns (x,y) and the variable of interest (Z) from the excel spreadsheet and generating contour plots and heat maps. However, I'm currently getting this error message with the following code.

"Error using meshgrid (line 59)

Subscripting into a table using one subscript (as in t(i)) is not supported. Specify a row

subscript and a variable subscript, as in t(rows,vars). To select variables, use t(:,i) or for

one variable t.(i). To select rows, use t(i,:).

Error in Specimen_84_HeatMap (line 5)

[X,Y] = meshgrid(x,y); "

Specimen_84 = readtable ('Specimen_84.xlsx');

x = Specimen_84(:,17);
y = Specimen_84(:,18);
[X,Y] = meshgrid(x,y);

Z = Specimen_84(:,11);

figure
contour (X,Y,Z)

Answer 1

[Walter Roberson]

Specimen_84 = readtable ('Specimen_84.xlsx');

x = Specimen_84{:,17};
y = Specimen_84{:,18};
[X,Y] = meshgrid(x,y);

Z = Specimen_84{:,11};

figure
contour (X,Y,Z)

However, this is not going to work. You are extracting as many Z as you have X or Y, but when you meshgrid you are creating an array which is (number of x) by (number of y), and contour would expect Z to be the same size.

You probably need to to use the X, Y, Z information as inputs to a scattered interpolant.

You could use https://www.mathworks.com/matlabcentral/fileexchange/38858-contour-plot-for-scattered-data or you could use

N = 50;    %adjust as desired. This is the resolution.

Specimen_84 = readtable ('Specimen_84.xlsx');

x = Specimen_84{:,17};
y = Specimen_84{:,18};
z = Specimen_84{:,11};

[X,Y] = meshgrid(linspace(min(x), max(x), N), linspace(min(y),max(y),N));
F = scatteredInterpolant(x, y, z);
Z = F(X, Y);

figure
contour(X,Y,Z)

[Shao Yang Zhang]

Hi! Walter Robertson, thank you for your answer. Your tip has helped tremendously and generated a graph. In my data set, there's 50 points each for X, Y, and Z so I wasn't sure why meshgrid cannot be performed, even though you mentioned that the arrays for x & y shared the same size as Z. I'm not too familiar about the scattered interpolant nor the tricontour function. What's your recommendation? My aims for this part of the project is to perform grid analysis on the XPM files, displayed originally as a heat map, in order to retrieve additional information.

[Walter Roberson]

There are some cases where data in columns is gridded data that has been "unrolled", such as

X    Y    Z
1    1    3
2    1    7
3    1    4
4    1   -1
1    2    0
2    2    8
3    2    4
4    2    0

instead of storing

    X        Y       Z
  1   1    1  2    3   0
  2   2    1  2    7   8
  3   3    1  2    4   4
  4   4    1  2   -1   0

That can happen. But when it does happen, you would not use meshgrid the way you did. You would instead need to extract the unique coordinates (1 2 3 4 for x, 1 2 for y) and meshgrid those, if you wanted to meshgrid... but more often you would just reshape() the vector of values (and possibly transpose) to create an appropriate grid of values.

If your data falls on a natural grid: reshape()

But... if your data does not fall onto a natural grid, then what you have instead is scattered points around the surface, and you want to use the hints they provide to deduce a surface. This is the case you would use scatteredInterpolant for: your input is scattered around, and you want to estimate what the surface might look like, and draw contours appropriately.

The file exchange contribution says that "Contours are computed directly from the triangulation rather than interpolating back to a cartesian grid using GRIDDATA." In cases where you have reason to expect that the triangulation is representative of the surface, then the output of that approach can be more appropriate than using scatteredInterpolant. The differences have to do with deciding which points influence the output calculations and to what extent. You could try both and see which you prefer.

At the moment I can only find reference to XPM files that are an obsolete bitmap file; I doubt that is what you are using.

[Shao Yang Zhang]

Thank you for your response. I'm quite overwhelmed by all the information povided from the last post and would like some clarification:

1. How can I tell if my data is rolled or unrolled? Should I navigate to workspace to determine the dataset organization?
2. When ran the code, the image I got was contoured lines (Figure Attached). Does that indicate the data fell into a natural grid?
3. When referring to estimating the working surface and drawing appropriate contour, can you elaborate on how I can accomplish this aspect? The contour plot did not fill the graph's window and lines were predetermined. It would be beneficial to know how I can alter the window size for the plot to fill the window and determine the depth of each contour line.

N = 50;    %adjust as desired. This is the resolution.

Specimen_84 = readtable ('Specimen_84.xlsx');
x = Specimen_84{:,17};
y = Specimen_84{:,18};
z = Specimen_84{:,11};

[X,Y] = meshgrid(linspace(min(x), max(x), N), linspace(min(y),max(y),N));
F = scatteredInterpolant(x, y, z);
Z = F(X, Y);

figure
contour(X,Y,Z);
colorbar;

[Walter Roberson]

Try this:

nux = length(uniquetol(x));
nuy = length(uniquetol(y));
if nux * nuy == numel(z)
    fprintf('z is probably a natural grid\n');
else
    fprintf('z is probably NOT a natural grid\n')
end