Creating a series of x-y plots

Hey, I want to make a set of x-y plots based on data with the following format:

"Tag    Property-A    Property-B
X1        A11           B11    
X1        A12           B12
...........................
X2        A21           B21
..........................."

Properties A and B are the x and y variables. Each tag is matched with several pairs of A and B, and each plot should include data with a specific tag. Since there are a good many tags, it is impossible to manually separate the data. So is there any commands in Matlab that can be used to create such plots? I hope I have clearly stated my question. Thanks. Lin

P.S. I just realized the format couldn't display properly, so I'll briefly describe the data: there are three columns of data: Tag, Property A, Property B. Each tag encompasses a certain amount of x-y pairs, and the purpose is to make every one of the tags a plot showing the scattered x-y relationship.

A sample data file is attached.

[Chad Greene]

Can you give an example of your data, so we can have some numbers and strings to play with?

[Lin]

Yes - the sample data are posted as an attachment. Thanks.

[Image Analyst]

I fixed the formatting for you, but please read this http://www.mathworks.com/matlabcentral/answers/13205-tutorial-how-to-format-your-question-with-markup

[Lin]

Thanks. This is helpful. I'm still trying to get a hang of the formatting rules here.

Answer 1

[Chad Greene]

If you have three columns, call them station, depth, and salinity, you could plot all the station 1 data as red pentograms, station 2 as blue squares, etc, like this:

plot(salinity(station==1),depth(station==1),'rp-'); 
hold on
plot(salinity(station==2),depth(station==2),'bs-'); 
plot(salinity(station==3),depth(station==3),'mo-'); 
plot(salinity(station==4),depth(station==4),'k^-'); 
plot(salinity(station==5),depth(station==5),'gx-'); 

legend('station 1','station 2','station 3','station 4',...
    'station 5','location','southwest')
legend boxoff 

box off
ylabel('depth (m)')
xlabel('salinity (psu??)') 
set(gca,'ydir','reverse')
axis tight

[Chad Greene]

oops, I tinkered a little bit and messed up the legend before I saved the figure. That's why the legend doesn't match the data.

[Lin]

Hey Chad, Thank you for the nice demonstration! It looks good to make all stations on a same plot when there are fewer stations. But it might be a bit of a problem of doing so for 70+ cruise stations, in which circumstance giving each station (or a couple of stations) their own plot will be needed. And I see there doesn't seem to be a solution to expect Matlab to collect data of a specific station and create plots automatically, as opposed to manually inputting station # in the commands. But 70 stations are not too bad to cope with, though typing may actually take quite a while. Anyway, thanks for spending time looking at this! Regards, Lin

[Chad Greene]

Don't manually type plot lines 70 stations! If you want to do all of them on the same plot, this would work:

stationColors = jet(70); % creates 70 colors; 
hold on; 

for k = 1:70
    plot(salinity(station==k),depth(station==k),...
        '.-','color',stationColors(k,:)); 
end
box off
ylabel('depth (m)')
xlabel('salinity (psu??)') 
set(gca,'ydir','reverse')
axis tight

Similarly, you could skip the loop and use scatter. You could even make the colors reflect something physically meaningful, rather than simply the station number. The profiles could be colored by latitude, or distance from the center of some gyre, or whatever interesting variable you're investigating.

If you only want to plot a single station, say, station number 15, you could use the example above like this:

k = 15; % <-- enter station number here
plot(salinity(station==k),depth(station==k),...
    '.-','color',stationColors(k,:)); 
box off
ylabel('depth (m)')
xlabel('salinity (psu??)') 
set(gca,'ydir','reverse')
axis tight

[Lin]

Thanks dude! I guess I can pretty much figure it out now. REALLY appreciate it!

Answer 2

[Image Analyst]

Just use readtable

t = readtable(xlsFullFileName);

Then extract the columns, use unique() to find unique tag names, then use logical indexing in a loop to extract rows for each tag in turn and plot. Let us know if you can't figure it out. Sorry, but I have to run off somewhere now.

[Chad Greene]

Just a note, I think readtable requires version 2013b or later.

[Lin]

Thank you for your reply. By "... use unique() to find unique tag names...", do you mean that each tag needs to be manually quoted for the related x-y pairs to be extracted? There are actually too many tags to be searched for one by one. Is there any way the program can perform such a loop as to go through the Tag column, extract the x-y pairs for an identical tag, and keep the loop going from a different tag?

[Image Analyst]

If you have tags, like "X1", "X2", "ABC", "This stuff", "Tag 42", whatever, and some of them occur just once, and some of them multiple times, you can get a list of all the names, just listed once, with the unique() function.

uniqueNames = unique(tagList);

Once you have that you can get a list of all rows that have a particular tag, say "X1", like rows 1, 2, 14, 33, and 73 have tag "X1". Then you can get columns 2 and 3 for just those rows and do whatever you want with them.

[Lin]

OK, I think I get the point. I may need to dig it a little more later on, as I haven't figured out the commands to query the columns with a certain tag. But it's OK since I've done with the work by doing it manually. Thanks for the information anyway.

[Image Analyst]

Use ismember().