How to clear array in between multiple plots in for loop?

Hi everyone,

I'm having some trouble doing multiple plots in a for loop. The arrays I use for my plot, stores data from previous iteration, messing up the plots.

Based on a .csv file containing data from 20+ devices, I wan't to seperate the data unique to each device and plot it. I do this as follows.

 fid = fopen('Regular_events_24_10.csv');
 readData = textscan(fid, '%f %s %q %f %f %f %s %s %s %s %s %s %s %s %s %s', 'Headerlines', 1, 'Delimiter', ',');
 device_name = ["device_name1", "device_name2"....]

 for u = 1:length(device_name)
    k = 1;
    for n = 1:length(readData{:,2})
        if strcmp(strcat('"',device_name(u),'"'),readData{1,2} (n,1))
            %disp(n);
            for i = 1:16
                temp_array{1,i} (k,1) = (readData{1,i} (n,1));
            end
            k = k + 1;
            %disp(k);
        end
    end

 device_time = temp_array{1,3} (:,1);
 state = temp_array{1,4} (:,1);
 progress = temp_array{1,6} (:,1);
 desk_setting = temp_array{1,7} (:,1);
 user_presence = temp_array{1,8} (:,1);
 desk_height = temp_array{1,9} (:,1);
 IR_obj = temp_array{1,12} (:,1);

 figure('rend','painters','pos',[150 60 1100 750])

 subplot(2,1,1);
 hold on
 title(device_title(u));
 plot(dateAndTimeSerial,desk_height, 'r');
 plot(dateAndTimeSerial,progress, 'k');
 plot(dateAndTimeSerial,desk_setting_int, 'b');
 datetick('x','HHMM');
 legend('desk height','progress','desk setting int', 'Location','west');
 hold off

 subplot(2,1,2);
 hold on
 plot(dateAndTimeSerial,IR_obj, 'r');
 plot(dateAndTimeSerial,user_presence_int, 'k');
 plot(dateAndTimeSerial,state, 'b');
 datetick('x','HHMM');
 legend('IR obj', 'user presence int', 'state', 'Location','west');
 hold off

 end

I have omitted some of the code, where I typecast some of the data and so on. Should be irrelevant for my question.

I have tried to clear and overwrite all arrays at the end of the for loop, but I doesn't help anything.

Hope you can help me out,

Best regards Mads

For reference, this is what a failed plot looks like.

[KL]

What's the structure of your csv file? It looks like you can do the whole process a lot simpler, data extraction, plotting and everything. Why don't you attach a sample csv file with similar structure?

[Mads Svendsen]

Hey KL,

The csv file looks like this:

 ID,"device_id","device_time","state","goal","progress","desk_setting","user_pressence","desk_height","BMP_pressure","BMP_temp","IR_obj","IR_amb","firmware_version","bluemix_id","bluemix_type"
 212490,"Shifty_23","2017-10-24 10:50:10",3,3,0,False,True,"71.1","1017.334424","33.89","26.37","30.07",3,"Shifty_23","Shifty_bot"
 212489,"Shifty_20","2017-10-24 10:48:58",8,3,0,False,False,"67.7","1023.169846","35.3","21.71","30.83",3,"Shifty_20","Shifty_bot"
 212488,"Shifty_7","2017-10-24 10:48:56",8,3,0,False,False,"65.7","1023.018408","33.89","23.27","30.95",3,"Shifty_7","Shifty_bot"
 212487,"Shifty_19","2017-10-24 10:48:52",8,3,0,False,False,"67.3","1022.899485","31.7","22.41","28.35",3,"Shifty_19","Shifty_bot"
 212486,"Shifty_6","2017-10-24 10:48:50",8,3,0,False,False,"77.2","1023.477279","32.08","23.35","29.05",3,"Shifty_6","Shifty_bot"
 212485,"Shifty_17","2017-10-24 10:48:47",8,3,0,False,False,"69","1022.936831","32.17","22.25","28.37",3,"Shifty_17","Shifty_bot"

I'm not using Matlab very often, so I won't be suprised if there's an easier and simpler way ;)

Best regards Mads

Answer 1

[Guillaume]

The immediate problem is that indeed you fail to clear your temp_array at the beginning of each step of your u loop. Simply fixed with

for u = 1:length(device_name)
   temp_array = {};
   %...

would fix that. Saying that, your code is very inefficient. You do need to learn to vectorise your code rather than looping over everything. Your n loop can be easily replaced. The i loop as well with a bit more effort. As a bonus, it wouldn't require to clear the temp_array as it would recreate a brand new one each time:

for u = 1:numel(device_name) 
  isdevice = strcmp(['"', device_name(u), '"'], readData{2});  %no need to index you can compare the whole column at once
  temp_array = cellfun(@(col) col(isdevice), readdata, 'UniformOutput', false);
  %...

However, your life would be a lot easier if you used readtable to import your data. A table is a lot easier to manipulate that a cell array of columns.

readdata = readtable('Regular_events_24_10.csv');  
%import all done however we can clean up some columns
readdata.device_time = datetime(readdata.device_time);
readdata.desk_setting = strcmp(readata.desk_setting, 'True');
readdata.desk_setting = strcmp(readata.desk_setting, 'False');
%note that the above can be done by readtable if you override the import options. See the doc

Looping over the devices is trivial

for device = unique(readdata.device_id)'
  device_name = device{1};
  filtered_table = readdata(strcmp(readdata.device_id, device_name), :);
  figure('rend','painters','pos',[150 60 1100 750]);
  subplot(2,1,1);
  hold on
  title(device_name);
  plot(filtered_table.device_time, filtered_table.desk_height, 'r');
  plot(filtered_table.device_time, filtered_table.progress, 'k');
  plot(filtered_table.device_time, filtered_table.desk_setting, 'b');
  datetick('x','HHMM');
  legend('desk height','progress','desk setting', 'Location','west');
  hold off
  %...

[Mads Svendsen]

Thanks a lot Guillaume :)

It works perfectly and it's a lot simpler.