Error in datetime array format for writing new matrix

Hello matlab expert, I have this timeseries data in txt format. the first column contain year data in year decimal format. I want to convert the year decimal format into datetime and save it as new txt file but I got an error in writematrix because the datetime format. what function should I use?

clc;clear; close all;
format long g
filename1 = 'C:\ZTD\PWVctul.txt';
data1 = readmatrix(filename1);
T1 = array2table(data1);
tt = table2timetable(T1, 'RowTimes',datetime(0,1,1) + years(data1(:,1)));
tt4 = timetable2table (tt);
pw = table2array (tt4(1:end,"data12"));
time = table2array (tt4(1:end,"Time"));
newdata = [ty pw];
writematrix (newdata,'PWVctuldatetime.txt','Delimiter','space');

[dpb]

Put into a table and use writetable instead...

But, do you know the expected time to be returned for the given values?

> datetime(0,1,1)+years(2014.99980974125)       % the first time value in the file
ans = 
  datetime
   31-Dec-2014 13:37:56
>> 0.99980974125*365                            % comparison conversion; 2014 is not leap year
ans =
  364.9306
>> ans-364                                      % is Dec 31 but almost end of day
ans =
    0.9306
>> ans*24
ans =
   22.3333
>> 

Seems to me the datetime conversion isn't correct hour, even though it gets day.

Futzing with starting point in datetime changes the day, but it always seems to return the same time...

[Stephen23]

@dpb: the approach shown is fundamentally flawed, because:

• it uses a fixed-length year of exactly 365.2425 days, see https://www.mathworks.com/help/matlab/ref/duration.years.html
• it does not take into account the (accumulated) different year lengths since year 0.

This is easy to illustrate with a very simple example which should be exactly midnight on Jan 1st, but is not:

datetime(0,1,1) + years(2022)

ans = datetime
   31-Dec-2021 08:02:24

To get the correct year length for any specific year (required for that fraction), something like CALYEARS or DATESHIFT or two DATETIMES or similar would be required.

[dpb]

I figured it had to be somesuch -- but didn't spend much time thinking about it nor about that years is the average duration animal...I was aware of that, just didn't think about it at the time.

Seems like a reasonable enhancement to the conversion types in datetime to build into it.

[Stephen23]

"Seems like a reasonable enhancement to the conversion types in datetime to build into it."

I asked about this on another thread, got this response:

https://www.mathworks.com/matlabcentral/answers/1658145-decimail-years-date-to-juliandate#comment_2017165

Answer 1

[Stephen23]

Ah, fractional years. Here is one way to convert a fractional year (yes, even leap years) to datetime:

format long G
M = readmatrix('PWVctul.txt')

M = 1831×2

1.0e+00 * 2014.99980974125 0.0627371458432432 2015.00236052519 0.0619196013484259 2015.00509549932 0.0583826355657704 2015.00783049723 0.0549544140538488 2015.01057022325 0.0497257245907533 2015.01330994928 0.0412166574719457 2015.01604967531 0.0460983177325905 2015.01878464955 0.0448768178251868 2015.02151964734 0.0450682606219937 2015.02425937337 0.0464028118802565

X = M(:,2); % data
Y = M(:,1); % year
Z = datetime(floor(Y),1,1,'Format','y/MM/dd HH:mm:ss.SSSSSSSSS');
Z = Z + mod(Y,1).*((Z+calyears(1))-Z)

Z = 1831×1 datetime array
   2014/12/31 22:20:00.000057449
   2015/01/01 20:40:41.522390028
   2015/01/02 20:38:11.666557966
   2015/01/03 20:35:42.560648069
   2015/01/04 20:35:42.560410186
   2015/01/05 20:35:42.560494974
   2015/01/06 20:35:42.560572592
   2015/01/07 20:33:12.708211031
   2015/01/08 20:30:43.598515132
   2015/01/09 20:30:43.598592750
   2015/01/10 20:30:43.598677538
   2015/01/11 20:30:43.598762327
   2015/01/12 20:30:43.598524444
   2015/01/13 20:28:13.749941713
   2015/01/14 20:25:44.636782485
   2015/01/15 20:25:44.636544602
   2015/01/16 20:25:44.636629390
   2015/01/17 20:25:44.636714179
   2015/01/18 20:23:14.791594778
   2015/01/19 20:20:45.674649549
   2015/01/20 20:20:45.674734337
   2015/01/21 20:20:45.674811954
   2015/01/22 20:20:45.674896743
   2015/01/23 20:18:15.833247843
   2015/01/24 20:15:46.712832112
   2015/01/25 20:15:46.712916901
   2015/01/26 20:15:46.712686189
   2015/01/27 20:15:46.712763806
   2015/01/28 20:13:16.874900907
   2015/01/29 20:10:47.751014676

T = table(Z,X)

T = 1831×2 table
                  Z                          X         
    _____________________________    __________________

    2014/12/31 22:20:00.000057449    0.0627371458432432
    2015/01/01 20:40:41.522390028    0.0619196013484259
    2015/01/02 20:38:11.666557966    0.0583826355657704
    2015/01/03 20:35:42.560648069    0.0549544140538488
    2015/01/04 20:35:42.560410186    0.0497257245907533
    2015/01/05 20:35:42.560494974    0.0412166574719457
    2015/01/06 20:35:42.560572592    0.0460983177325905
    2015/01/07 20:33:12.708211031    0.0448768178251868
    2015/01/08 20:30:43.598515132    0.0450682606219937
    2015/01/09 20:30:43.598592750    0.0464028118802565
    2015/01/10 20:30:43.598677538    0.0451572386581995
    2015/01/11 20:30:43.598762327    0.0523973221610946
    2015/01/12 20:30:43.598524444    0.0631753909682948
    2015/01/13 20:28:13.749941713    0.0593898128672308
    2015/01/14 20:25:44.636782485    0.0557321130217992
    2015/01/15 20:25:44.636544602    0.0555264085669299

writetable(T,'newfile.txt') % no problem

[Anak Agung Adhi Dermawan]

Thank you sir

[Stephen23]

Another approach to converting from decimal years to datetime:

format long G
M = readmatrix('PWVctul.txt');
Y = M(:,1); % year
Z = datetime(floor(Y)+[0,1],1,1,'Format','y/MM/dd HH:mm:ss.SSSSSSSSS');
Z = Z(:,1) + mod(Y,1).*diff(Z,1,2)

Z = 1831×1 datetime array
   2014/12/31 22:20:00.000057449
   2015/01/01 20:40:41.522390028
   2015/01/02 20:38:11.666557966
   2015/01/03 20:35:42.560648069
   2015/01/04 20:35:42.560410186
   2015/01/05 20:35:42.560494974
   2015/01/06 20:35:42.560572592
   2015/01/07 20:33:12.708211031
   2015/01/08 20:30:43.598515132
   2015/01/09 20:30:43.598592750
   2015/01/10 20:30:43.598677538
   2015/01/11 20:30:43.598762327
   2015/01/12 20:30:43.598524444
   2015/01/13 20:28:13.749941713
   2015/01/14 20:25:44.636782485
   2015/01/15 20:25:44.636544602
   2015/01/16 20:25:44.636629390
   2015/01/17 20:25:44.636714179
   2015/01/18 20:23:14.791594778
   2015/01/19 20:20:45.674649549
   2015/01/20 20:20:45.674734337
   2015/01/21 20:20:45.674811954
   2015/01/22 20:20:45.674896743
   2015/01/23 20:18:15.833247843
   2015/01/24 20:15:46.712832112
   2015/01/25 20:15:46.712916901
   2015/01/26 20:15:46.712686189
   2015/01/27 20:15:46.712763806
   2015/01/28 20:13:16.874900907
   2015/01/29 20:10:47.751014676

... etc.