merge date time

How can I merge the two columns together (into one) and generate a date time number for data like

20101229  154907
20101229  154908
20101229  154908
20101229  154909
20101229  154909
20101230  93001
20101230  93001
20101230  93004

and a result like

2010229154907 for row 1
2010229154907 for row 2 and so on

dates and times are doubles

thanks

Answer 1

[Walter Roberson]

Do not accumarray() using these numbers as the subscripts. Instead,

Mu = unique(mergedtimes);
[tf, Muidx] = ismember(mergedtimes, Mu);

Then uses MuIdx(:) as the subscript for accumarray() purposes. Afterwards, you can map back row K of your accumarray results to time value Mu(K)

[Tiina]

thanks Walter that works well without memory issues ..

one question though, when I merge times and perform operations i lose some rows in the merged times column as a result of the operations.

I wish to demerge what remains in the merged times columns and get back the dates and times split again.

should i covert to string and read by character or is there a way that works on double?

[Walter Roberson]

mod(mergedtime, 1000000) to get the time data,

fix(mergedtime ./ 1000000) to get the data data.

[Tiina]

yea awesome

Answer 2

[Walter Roberson]

mergedtimes = data(:,1) * 1000000 + data(:,2);

Answer 3

[James Tursa]

If X is your original M x 2 data:

X(:,1)*1000000 + X(:,2)

[Tiina]

I tried before posting a q and bec I am using accumarray at one point in the code am running out of memory. When I did not merge the columns (that is accumarray by either columns) i didn not run out of memory so i though that merging by multiplying by a number would be related to the issue ..

is there any tip for this case?