Generate bar graph in a for loop

So I have data in a 32*1 cell and I would like to plot them on bar graphs. The complicated part is I dont need 32 bar graphs, I would like to group these values in groups of 4. So I would have a total of 8 bar graphs. please help

clc
clear
warning('off','all')

d = dir('*.csv'); 
dm = dir('*.mat');
nm = numel(dm);
n = numel(d);
firewall = cell(32, 1);
firewall_array = cell(32, 1);
Y_Firewall_avg_4_24 = cell(32, 1);
UDT = cell(32, 1);
UDT_array = cell(32, 1);
X_UDT = cell(32, 1);
Y_UDT = cell(32, 1);
 Y_UDT_avg_4_24 = cell(32, 1);
Wan = cell(32, 1);
Wan_array = cell(32, 1);
X_Wan = cell(32, 1);
Y_Wan = cell(32, 1);
Y_Wan_avg_4_24 = cell(32, 1);
X_Firewall = cell(32, 1);
Y_Firewall = cell(32, 1);
Digitized_graphs = cell(32,1);
XDS = cell(32,1);
YDS = cell(32,1);
X_DS = cell(32,1);
Y_DS = cell(32,1);
Y_DS_avg_4_24  = cell(32,1);
y_Bar_4_24  = cell(32,1);
% X = cell(32,1);
% b = cell(32,1);
for k = 1 : 32 
    firewall{k} = readtable(sprintf('Scenario%d.csv', k));
    UDT{k} = readtable(sprintf('udt_scenario_%d.csv',k));
    Wan{k} = readtable(sprintf('WAN_Scenario%d.csv',k));
    
    firewall_array{k} = table2array(firewall{k}(:,[3,5]));
    UDT_array{k} = table2array(UDT{k}(:,[3,4]));
    Wan_array{k} = table2array(Wan{k});
   
    Y_Firewall{k} = [firewall_array{k}(:,1)]; 
    Y_UDT{k} = [UDT_array{k}(:,2)];
    Y_Wan{k} = [Wan_array{k}(:,2)]; 

    Y_Firewall_avg_4_24{k} = mean(Y_Firewall{k});
     Y_Wan_avg_4_24{k} = mean(Y_Wan{k});
    Y_DS_avg_4_24{k} = mean(Y_DS{k});
    Y_UDT_avg_4_24{k} = mean(Y_UDT{k});

bar([Y_Firewall_avg_4_24{k} Y_Wan_avg_4_24{k} Y_DS_avg_4_24{k}  Y_UDT_avg_4_24{k} ; Y_Firewall_avg_4_24{k+1} Y_Wan_avg_4_24{k+1} Y_DS_avg_4_24{k+1}  Y_UDT_avg_4_24{k+1} ])

end

[Adam Danz]

How does this quesiton differ from the one you asked yesterday?

https://www.mathworks.com/matlabcentral/answers/561158-bar-graph-in-a-for-loop

[JC]

More information. my question wasnt really answered yesterday

[Adam Danz]

Then you should have continued the conversation on that thread instead of deleting it.

Several volunteers already invested time to help you on that thread. They asked for more information and it wasn't provided. You ignored those volunteers, some of whom were very experienced, and decided to start a new thread to pull in more volunteers to help you. Your previous question could have been answered if you provided the information the volunteers requested. But since you deleted that question and the work that many people already did for you, you're back to square one.

[JC]

I am sorry this upsets you. I apoligize for waisting their time. I delted it to not have duplicates. This post offers more information. I apreciate your proffessional response to my mistake

[dpb]

What do you mean precisely by " to group these values in groups of 4. "?

You have stuffed tables and other things into cell arrays and we can't see inside those to have any idea what the data might actually be...

Think you'll need to attach a sample dataset and perhaps a sketch or more complete description of what you envision the resulting plots to look like--we just can't tell enough about the data to do anything from this.

I didn't see the other thread but I'd suspect the response there was the same one -- we need real data!

[JC]

So for example each 32 by 1 cell includes average values for 32 var values. They all look this way. If these were not in this cell I would have them named var1 through var32. There are 4 other 32 by 1 cells just like this but the var values are different. I would like to have 8 bar graph plots.

for example bar(x, [ a1 b1 c1 d1 ; a2 b2 c2 d2 ; a3 b3 c3 d3 ; a4 b4 c4 d4]);

However isntead of doing this number by number I wish to have it in a four loop. Please let me know if this does not make sense?

[dpb]

"32 by 1 cell includes average values for 32 var values. They all look this way. If these were not in this cell I would have them named var1 through var32. "

There's the root of your problem -- don't do that!!! Use a regular array and just keep track of "who's who in the zoo!" by index.

Then you can just use column indexing to address whichever columns you wish and build graphs or do whatever.

It doesn't make any sense to use readtable and then immediately just turn around and turn that table into an array -- and it doesn't make easy to write code generically when create multiple variables for related data that should be in one or two (or maybe four at the outside) arrays.

[JC]

So what do you suggest? How can I load the CSV files and then assign them variables and later use them for graphing? I only turned some into an array so I could subtract and multiply. Can you rad CSV files straight to an array?

[dpb]

Of course you can read directly into array -- if the data are just numeric. We don't know precisely what all these other files are.

And, you can operate on the data in a table without taking it out of the table as well...

What makes most sense depends on just what the data really are and the overall objectives...

[JC]

So the data in the csv files are numeric with the exception of the headers. I really want to take the averages and max values and put them on a bar graph. Reguardless of what the data is, how do you create a bar graph in a for loop where I can have it look similar to this?

[dpb]

Sure, that's also easy if you create arrays instead of individual variables. In Matlab, arrays are the ticket to almost all effective use of the product.

Not having your raw data, and being too lazy to try to build a set I just took your above--

dat=[121 210 449;454 216 500;420,197,454;436 204 512];      % some sample data
sys=categorical({'Firewall','Jones WAN','Server','WAN'});   % categorical array for x axis
hBar=bar(sys,dat);                                          % grouped bar
hTxt=arrayfun(@(hB) text(hB.XEndPoints,hB.YEndPoints,compose('%d',hB.YEndPoints), ...
    'HorizontalAlignment','center','VerticalAlignment','bottom'),hBar,'UniformOutput',false);
hLg=legend(compose('Scenario %d',1:3),'Location','southoutside','Orientation','horizontal');

yields

I'd go back to the reading portion, read each of the raw data files into a large array, the exact form of which would pick based on knowing more about the file content/organization than presently do -- but statstics can be computed by grouping variables or by referencing columns by ranges or overall either by row or by column depending on the organization.

You just need to pick the orientation so that you can generate the summary statistics as arranged above -- groups are rows, columns are variables in group. So the above is four rows by three columns.

We're fortunate if you're also using at least R2019b; prior releases did not expose the X/YEndPoints properties used above -- you had to use an undocumented property whose name I could never recall...have to give kudos to TMW altho it's been 20 years in the making.

Now if they would introduce a 'BarLabel' property to set with all it's associated properties instead of having to write them explicitly...

[JC]

Ok so now my question is. I want 8 of these graphs each looking like the one above with the eception of having one more scenario added. This is why I want to use a four loop. To produce several of these graphs untill the last bar graph represents scenario 28-scenario 32.

[dpb]

Again, start by building the full data set into an array (or at least no more than one array per scenario) -- then you can easily index the subsets (columns) desired to both calculate the statistics and do the plots.

If you'll attach a couple sample datasets to see what really look like, I'll try to look at it some more depth tonight if you still can't see your way through the process...

[JC]

So I would like to graph the averages and max values serperately. Ive spent hours playing with this and its driving me insane

[dpb]

Attach the .csv text files w/ the paperclip -- nobody can do anything with images.

I looked at your original code some more -- there are varying numbers of columns in the different files -- when read w/ readtable you'll get variable names -- is there some list of "who's who in the zoo" as to which should be selected from which file for the purpose of the graphs?

Need to write down the specific ground rules for what is wanted to be done so can write code to match. That's what you need to be able to define which columns count for what.

Ideal would be to zip up the lot so somebody if gets chance can actually solve your problem...

I can see enough to outline code as said but can't test much although the answer to the above Q? regarding the rules on how to know which columns to use for each is important piece of the puzzle.

Remember, we only know what you tell us; we can't know what you know about the problem so how to know what it makes sense to compare is all greek to us...

[dpb]

Is the above image an output table/spreadsheet of computed results or the input files? I'm guessing the former as that doesn't really match up with the initial code...but it doesn't seem to make a lot of sense in comparison either with the multiple columns.

I'd be glad to help further if could just figure out what the first-source data really are we're starting from...