Expected timetable data to be numeric.
Show older comments
I am trying to do a sftf(x) function, I have compiled the data and datetime has been combined using:
data = 'TA';
for date = datetime(date, 'InputFormat', 'dd-MM-yyyy hh:mm:ss')
end
I keep getting the error Error using signal.internal.utilities.validateattributesTimetable>localCheckAttribute
Expected timetable data to be numeric.
I am unsure where I'm going wrong. Is the imput of date/time incorrect?
Would using a DateVector be useful? I have 57001 rows in each variable (several variables per brain area, and 12 brain areas), it will be impossible to type it all out as e.g. [2014 10 24 12 45 07]. I have no idea how to do it for all rows.
2 Comments
It is unclear what this code is supposed to do:
data = 'TA';
for date = datetime(date, 'InputFormat', 'dd-MM-yyyy hh:mm:ss')
end
Why write a loop with nothing in it? Why define DATA but not use it? How does this relate to your screenshot?
"I am unsure where I'm going wrong. Is the imput of date/time incorrect?"
We don't know either, because we have no idea what DATE is. Nore what TT is in your screenshot.
Your screenshots show that you apparently call some functions, but we have no idea what is in those functions.
If you want help debugging this please upload your code files by clicking the paperclip button.
Antoinette Burger
on 16 Feb 2023
Edited: Antoinette Burger
on 16 Feb 2023
Accepted Answer
Star Strider
on 16 Feb 2023
Edited: Star Strider
on 16 Feb 2023
It would help to have the data.
I am not certain how the ‘date’ and ‘time’ values are being imported.
One possibility:
DateTime = datetime(TT.date, 'InputFormat','dd/MM/yyyy') + timeofday(datetime(TT.time, 'InputFormat','HH.mm.ss'))
It would also be helpful to know what the ‘time’ values represent. Something like this may be necessary —
time = {'14.11.0020'; '14.11.0021'}
HHmmssSS = cellfun(@(x){x(end-9:end-8) x(end-6:end-5) x(end-3:end-2) x(end-1:end)},time, 'Unif',0);
timec = cellfun(@(x)sprintf('%02d:%02d:%02d.%02d',str2double(x)),HHmmssSS, 'Unif',0);
T = datetime(timec, 'InputFormat','HH:mm:ss.SS', 'Format','HH:mm:ss.SS')
For whatever reason, 'InputFormat','HH.mm.ssSS' fails here.
EDIT — (16 Feb 2023 at 18:27)
Corrected typographical errors. Code unchanged.
.
17 Comments
Antoinette Burger
on 16 Feb 2023
Antoinette Burger
on 16 Feb 2023
I need to create a version of your timetable to demonstrate the correct coding.
It would actually be —
date = datetime({'18/01/2021'; '18/01/2021'}, 'Format','dd/MM/yyyy');
time = {'14.11.0020'; '14.11.0021'};
LACIN1 = rand(2,1);
T = table(date,time,LACIN1);
TT = table2timetable(T)
T = timetable2table(TT); % Temp[orarily Conmvert To 'table'
HHmmssSS = cellfun(@(x){x(end-9:end-8) x(end-6:end-5) x(end-3:end-2) x(end-1:end)},TT.time, 'Unif',0);
timec = cellfun(@(x)sprintf('%02d:%02d:%02d.%02d',str2double(x)),HHmmssSS, 'Unif',0);
DateTime = T.date + timeofday(datetime(timec, 'InputFormat','HH:mm:ss.SS', 'Format','HH:mm:ss.SS'));
DateTime.Format = 'dd/MM/yyyy HH:mm:ss.SS'
T = addvars(T,DateTime,'Before','date') % Add: 'DateTime'
TT = table2timetable(T) % Convert Back To 'timetable'
VN = TT.Properties.VariableNames
That should work.
.
Antoinette Burger
on 16 Feb 2023
Star Strider
on 16 Feb 2023
As always, my pleasure!
Antoinette Burger
on 16 Feb 2023
Edited: Antoinette Burger
on 16 Feb 2023
Walter Roberson
on 16 Feb 2023
Could you show a few of the entries from timec ? The first 3 perhaps? And verify that
datetime(timec(1:3), 'InputFormat','HH:mm:ss.SS', 'Format','HH:mm:ss.SS')
does indeed fail ?
Antoinette Burger
on 16 Feb 2023
Walter Roberson
on 16 Feb 2023
Please show us timec(1:3)
Antoinette Burger
on 16 Feb 2023
Antoinette Burger
on 16 Feb 2023
Edited: Antoinette Burger
on 16 Feb 2023
Walter Roberson
on 16 Feb 2023
Notice that in his demonstration, Star Strider used a different line of code,
HHmmssSS = cellfun(@(x){x(end-9:end-8) x(end-6:end-5) x(end-3:end-2) x(end-1:end)},TT.time, 'Unif',0);
That code applies to all rows in the table.
Please show us
celldisp(HHmmssSS(1:3))
celldisp(timec(1:3))
Star Strider
on 16 Feb 2023
@Antoinette Burger — I assumed ‘time’ is a cell array because I cannot imagine it being anything else and still having that format.
Can you share a few rows of your Excel file? That should be all that is necessary.
One way to do that would be to save it as a slightly different name to crreate a copy of it (the original file will remain unchanged) and then delete all but five or ten rows of the copied and renamed file. Alternatively, just copy the first few rows of the file and save it in a different name. (I am not certain which would be easiest.)
Then, upload the shortened copy here.
Antoinette Burger
on 16 Feb 2023
Antoinette Burger
on 16 Feb 2023
filename = 'https://www.mathworks.com/matlabcentral/answers/uploaded_files/1298125/sample_left_cing_table.csv';
T = readtable(filename)
class(T.time)
class(T.time(1))
T.time(1)
T.time(1).Format
readtable() by default is interpreting the time column as dd.MM.uuuu not as text . You would have to break part the pieces and reconstruct them, or else you would have to use readtable options to tell it that the variable should be text.
options = detectImportOptions(filename);
options = setvartype(options, 'time', 'char');
T = readtable(filename, options)
class(T.time)
class(T.time(1))
T.time(1)
dt = datetime(T.time, 'InputFormat', 'HH.mm.ss');
dt = dt - dateshift(dt, 'start', 'day');
T.date = T.date + dt;
T.date.Format = 'yyyy-MM-dd HH:mm.ss'
Antoinette Burger
on 16 Feb 2023
More Answers (0)
Categories
Find more on Data Type Identification in Help Center and File Exchange
on 16 Feb 2023
on 16 Feb 2023
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
