Calculate standard deviation with different time interval
显示 更早的评论
Good day, everyone.
Dummy_Raw.xlsx is a spreadsheet that contains data for each minute of a 24-hour period. I would like to calculate the standard deviation of Data_C for every five minutes. I realised I should use a loop, however it's a little tricky because my data is in time format and in excel files. The end result is expected to be as in Dummy Expected.xlsx. The empty value in the excel files will be considered as null or 0. Meaning that at the end, it will be: 288×3 table (refer below).
I'm hoping the community can assist me with this because I usually do it manually and it takes a long time.
DAY TIME Data_Exp
___ _____ ______
1 00:00 NaN
1 00:05 NaN
1 00:10 NaN
1 00:15 NaN
1 00:20 NaN
...
1 17:25 0.010648
1 17:30 0.010981
1 17:35 NaN
1 17:40 0.013079
...
1 23:50 NaN
1 23:55 NaN
3 个评论
Dyuman Joshi
2022-2-15
编辑：Dyuman Joshi
2022-2-15
You can work it through a for loop. Can you show what you have tried?
Walter Roberson
2022-2-15
Have you tried converting to a timetable() object and using retime() ?
采纳的回答
Walter Roberson
2022-2-17
Here is a considerable improvement in performance:
Load the data
filename = "PRN1_365.xlsx";
opt = detectImportOptions(filename);
opt = setvartype(opt, "DAY", "double");
opt = setvartype(opt, "TIME", "duration");
opt = setvartype(opt, "TEC0", "double");
opt.SelectedVariableNames = ["DAY", "TIME", "TEC0"];
data = readtable(filename, opt);
% Create a datetime
data.TimeStamp = datetime(2014,01,01) + data.DAY-1 + data.TIME;
data.TimeStamp.Format = "MM-dd-yy HH:mm";
tic
% Convert the table to a timetable
dataTT = table2timetable(data(:,["TimeStamp", "TEC0"]),"RowTimes","TimeStamp");
toc
tic
% Retime for everyday with 5 minutes in 24 hrs
PRNTT5min = retime(dataTT(:,"TEC0"),"regular","mean","TimeStep",minutes(5));
toc
On my system, this executes in a fairly small number of seconds and does not run out of memory.
By contrast, before I added in the detectImportOptions, the code took several minutes and then MATLAB would abruptly quit -- probably ran out of memory. Because of all the empty rows of TEC0 data, MATLAB's default detection is that most columns are character.
3 个评论
Ann
2022-2-27
Hi Walter.
I appreciate your response, but I have a concern on something. The retime function (only for time) operates as predicted, with 5 minutes interval for every 24 hours in a 365-day period.
However, the TEC0 (or TECRate0) data is a bit puzzling. This is because the first photo displays data from 17:25-17:29 while the second photo, it's simply taking the value 17:26 and putting it at 17:25. Which supposedly PRNTT5min should be a standard deviation of TEC at 5 minutes. This confusion happened for all data. I hope you may assist on this.
Walter Roberson
2022-2-27
编辑：Walter Roberson
2022-2-27
Ah. At the location of 'mean' in the retime() command, replace the 'mean' with a handle to a function
@(x)std(x,[],1,'omitnan')
Note that for blocks that have no data in the time span, the result will be nan. Blocks that only have nan data within the time span will also generate nan. Blocks that have exactly one finite numeric reading will generate 0 no matter what the reading is. You will only get a nonzero result if there are multiple finite entries in a time slot (and, of course, they are not all the same)
Ann
2022-2-27
更多回答（0 个）
类别
在 帮助中心 和 File Exchange 中查找有关 Descriptive Statistics 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
Translated by 
