Error using datenum (line 201) DATENUM failed.

Help me!

I have this error, and I don't know why.

Error using datenum (line 201)
DATENUM failed.

Caused by:
Error using dtstr2dtnummx
Failed to convert from text to date number.

My script is this:

[ndata, text, alldata]=xlsread('positive_read_Copia.xlsx'); % inside there are data (dd/mm/yyyy hh:mm:ss) and coordinates
date=date.';
tmp=char(date);
tmp_date=datenum(tmp,'dd/mm/yyyy HH:MM');
DateVector = datevec(tmp_date);
Date_minute=(tmp_date-tmp_date(1))*24*60;   % Time, in minutes compared to the 1st georeferenced tweet
clear tmp;

[Cris LaPierre]

Please share your spreadsheet. You can attach it using the paperclip icon.

[Rachele Franceschini]

Done!

Answer 1

[Stephen23]

Do not use outdated, discouraged, deprecated DATENUM (or for that matter DATEVEC or DATESTR).

Import your data using READTABLE, which automatically identifies the dates and imports them as DATETIME objects:

T = readtable('positive_read_Copia.xlsx')

T = 1108×3 table
      tweet_created_at      Latitude    Longitude
    ____________________    ________    _________

    28-Sep-2020 01:13:00     4.8186       45.728 
    28-Sep-2020 02:11:00    -80.217       25.783 
    28-Sep-2020 03:20:00     4.4894       45.895 
    28-Sep-2020 08:38:00    -74.245       40.617 
    28-Sep-2020 09:32:00     4.0817       44.128 
    28-Sep-2020 09:46:00     4.0817       44.128 
    28-Sep-2020 10:20:00     103.45       35.845 
    28-Sep-2020 10:22:00     4.0817       44.128 
    28-Sep-2020 10:41:00     2.5067       49.098 
    28-Sep-2020 11:55:00     6.8992        49.53 
    28-Sep-2020 12:34:00     15.314      -4.3317 
    28-Sep-2020 12:50:00      35.25           32 
    28-Sep-2020 13:46:00     -14.75         16.5 
    28-Sep-2020 14:05:00          2           47 
    28-Sep-2020 14:36:00     7.4167       43.733 
    28-Sep-2020 16:24:00     9.2167       45.617 

Then your task is much easier too, simply subtract to get the differences as DURATION objects:

D = T.tweet_created_at - T.tweet_created_at(1)

D = 1108×1 duration array
   00:00:00
   00:58:00
   02:07:00
   07:25:00
   08:19:00
   08:33:00
   09:07:00
   09:09:00
   09:28:00
   10:42:00
   11:21:00
   11:37:00
   12:33:00
   12:52:00
   13:23:00
   15:11:00
   15:24:00
   15:42:00
   16:55:00
   18:10:00
   18:15:00
   21:16:00
   21:26:00
   22:42:00
   30:23:00
   31:14:00
   32:11:00
   32:23:00
   37:02:00
   37:55:00

which is then trivial to convert into MINUTES:

M = minutes(D)

M = 1108×1

0 58 127 445 499 513 547 549 568 642

[Rachele Franceschini]

but doing this I have another error :(

One moment, this is the script. In my opinion there is one problem with format of the datatime.

I don't know.

%% Loading dataset
    % Loading geolocalized tweets 
[ndata, text, alldata]=xlsread('positive_read_Copia.xlsx');

    % Loading GIS layer for the map
S = shaperead('Countries.shp');

for(i=1:(size(text,1)-1))
    date(i)=text(i+1,1);
end

X(:,1)=ndata(:,2);  % longitude
X(:,2)=ndata(:,1);  % latitude

date=date.';
tmp=char(date);
tmp_date=datenum(tmp,'dd/mm/yyyy HH:MM');
DateVector = datevec(tmp_date);
Date_minute=(tmp_date-tmp_date(1))*24*60;   % Time, in minutes compared to the 1st georeferenced tweet
clear tmp;

%% Run DBSCAN Clustering Algorithm
epsilon_time=60;    % Maximum duration (in minutes) to declare that 2 points are temporally correlated 
epsilon_dist=30;    % Maximum distance (in km) to declare that 2 points are spatially correlated 
MinPts=20;          % Minimun number of correlated points to declare a new cluster

IDX=DBSCAN_time(X,Date_minute,epsilon_dist,epsilon_time,MinPts);

%% Dimension reduction (ie. groupings  duplicate points having  the same geographic coordinates)
tmp_IDX=X;
tmp_IDX(:,3)=IDX(:);
[C,ia,ic] = unique(tmp_IDX(:,1:3),'rows');
a_counts = accumarray(ic,1);
IDX_points2point = [C, a_counts];

%% Plot Results
    
subplot(3,1,[1,2])
mapshow(S);
hold on

Legends=PlotClusterinResult_points2point_floods(X, IDX_points2point);
Legends_histo = {};
title(['DBSCAN Clustering (\epsilon distance = ' num2str(epsilon_dist) ' km, \epsilon time = ' num2str(epsilon_time) ' minutes, MinPts = ' num2str(MinPts) ')']);xlim([-6 10])
ylim([41 52])
k=max(IDX);
Colors=hsv(k);
subplot(3,1,[1,2])

if k>=0
    for(i=1:k)
        tmp_cluster=find(IDX==i);
        tmp_bd = [X(tmp_cluster(:),1) X(tmp_cluster(:),2)];
        bound=boundary(tmp_bd(:,1),tmp_bd(:,2));
        tmp_histo=tmp_date(tmp_cluster(:));
        stat_clust(i,1)=mean(tmp_histo*24*60)/(24*60);
        stat_clust(i,2)=std(tmp_histo);
        Color = Colors(i,:);
        subplot(3,1,[1,2])
        hold on
        plot(tmp_bd(bound,1),tmp_bd(bound,2),'Color',Color);
        Legends{end+1} = ['Area cluster n°' num2str(i)];
        subplot(3,1,3)
        hold on
        grid on
        h=histogram(tmp_histo);
        h.FaceColor = Color;
        h.BinWidth = 8/24;
        Legends_histo{end+1} = ['Cluster n°' num2str(i),' Tmean=' datestr((stat_clust(i,1)),'dd-mm HH'),'+/-' datestr((stat_clust(i,2)),'HH'),' hours'];
        clear tmp_cluster tmp_bd bound tmp_histo
    end
end

subplot(3,1,3)
legend(Legends_histo);
legend('Location', 'NorthEastOutside');
xlabel('Date')
ylabel('Distribution')
axis([datenum('2020-09-30') datenum('2020-10-07') 0 200])
datetick('x','dd-mm HH')

[Stephen23]

The code that you show in your comment is completly unrelated to my answer, so it is not clear what you expect from me. In any case, you should replace those deprecated functions with DATETIME and DURATION objects.

"but doing this I have another error"

Please show the complete error message. This means all of the red text.

[Sean de Wolski]

You'll also want to look at https://www.mathworks.com/help/releases/R2021b/stats/dbscan.html and either geobubble or geoplot https://www.mathworks.com/help/releases/R2021b/matlab/ref/geoplot.html.

[Rachele Franceschini]

Error is: I have the map and the graph empty.