Help extracting rows based on unique column 1 data

2 次查看（过去 30 天）
Brantosaurus
Brantosaurus 2022-11-19
回答： Campion Loong 2022-11-30
I have a large MxN matrix of data from which i would like to extract specific rows of data.
The first column of the matrix contains a chronological timebase in seconds, in ascending order.
t1 a1 b1 .. n1
t2 a2 b2 .. n2
:
talpha ..
:
tbeta ..
:
tgamma ..
:
tm am bm .. nm
I know the unique times of 3 events (talpha, tbeta, tgamma) that occur in column 1.
How do i extract the rows in chronological order based on the knowledge of their unique times?
Tried categorical with valueset and ordinal but no luck!
  2 个评论
Image Analyst
Image Analyst 2022-11-19
How large is large? How many GB or rows? If it really is large, like hundreds of millions of rows or several GB of data then you may need to use memmapfile. If you just have a few million rows or less, that's not large.
Brantosaurus
Brantosaurus 2022-11-19
移动：Voss 2022-11-19
Not THAT large!!!
Maybe 10,000 x 20 at most

请先登录，再进行评论。

请先登录，再回答此问题。

采纳的回答

Star Strider
Star Strider 2022-11-19
There is a lot I don’t unerstand about this problem.
However, if you want to find the unique occurrences of each of the ‘talpha’ , ‘tbeta’ and ‘tgamma’ so that they return the indices of the first incidence of each value, use the unique function with the first two outputs (the unique values, and the indices of the first instance of each value) with the 'stable' argument (to prvent sorting the result).
  2 个评论
Brantosaurus
Brantosaurus 2022-11-22
Apologies for the vagueness. Had a play around and came up with something following your excelent advice. Thanks for mentioning the unique function and indices. Not heard of those before. Best regards.

请先登录，再进行评论。

更多回答（1 个）

Campion Loong
Campion Loong 2022-11-30
Ran in:
I would first put my data in a timetable. It is specifically built for time based data like yours.
% Make some pretend data for your a#/b#/c# numbers
A = rand(10000,1); B = rand(10000,1); C = rand(10000,1);
% Use timetable. Here I assume regular timesteps. If not, use the
% 'RowTimes' N/V pair instead
tt = timetable(A, B, C, 'TimeStep', seconds(1))
tt = 10000×3 timetable
Time A B C ________ ________ _______ ________ 00:00:00 0.046518 0.25765 0.11216 00:00:01 0.021558 0.19586 0.078626 00:00:02 0.325 0.81709 0.9364 00:00:03 0.11313 0.10249 0.36468 00:00:04 0.65928 0.43818 0.4145 00:00:05 0.59172 0.30821 0.52174 00:00:06 0.39627 0.23423 0.42336 00:00:07 0.42042 0.62615 0.8902 00:00:08 0.62093 0.60716 0.098711 00:00:09 0.030026 0.18906 0.49313 00:00:10 0.5556 0.51185 0.6461 00:00:11 0.65488 0.38215 0.59968 00:00:12 0.48795 0.79178 0.69614 00:00:13 0.040898 0.99794 0.59517 00:00:14 0.17782 0.81708 0.12582 00:00:15 0.64031 0.78373 0.81385
If you know the exact time of your unique events
t_alpha = seconds(10);
t_beta = minutes(150);
t_gamma = hours(2);
% Index directly with time. Get all the rows in one go like below, or
% separately as your preferred:
tt([t_alpha; t_beta; t_gamma], :)
ans = 3×3 timetable
Time A B C ________ _______ _______ _______ 00:00:10 0.5556 0.51185 0.6461 02:30:00 0.44544 0.32722 0.23712 02:00:00 0.48217 0.21062 0.37704
Now, if you are not sure the 'exact' time (e.g. your data has some noise)
tt.Time = tt.Time + milliseconds(randn(10000,1)); % mix in some made-up noise
% Exact time index will turn up nothing ...
tt([t_alpha; t_beta; t_gamma], :)
ans = 0×3 empty timetable
% Use withtol to index with time tolerance:
wt = withtol([t_alpha; t_beta; t_gamma],seconds(0.5))
wt =
timetable withtol subscript: Select timetable rows matching the following times: 10 sec 9000 sec 7200 sec with tolerance of +/- 0.5 sec See Select Times in Timetable.
tt(wt, :)
ans = 3×3 timetable
Time A B C ________ _______ _______ _______ 00:00:10 0.5556 0.51185 0.6461 02:30:00 0.44544 0.32722 0.23712 02:00:00 0.48217 0.21062 0.37704

类别

在 Help Center 和 File Exchange 中查找有关 Data Preprocessing 的更多信息

标签

产品

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by