unique across tables within table

Question

Paolo Mazzoleni 2024-8-15

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2145419-unique-across-tables-within-table

评论： Paolo Mazzoleni 2024-8-29

Hello

I've got what in the screenshots below, so RawFileIndex is a table that contains Paramaters that is another table (or a 1x1 cell containing a table? I'm a bit confused...)

I want to know all the parameter ID accross all the files, something like unique(RawFileIndex.Parameters{:,1}.ID) but this is clearly not the right syntax... how do I do it?

thanks

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Stephen23 2024-8-15

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2145419-unique-across-tables-within-table#answer_1499209

在 MATLAB Online 中打开

U = unique(vertcat(RawFileIndex.Parameters{:}).ID)

5 个评论
显示 3更早的评论隐藏 3更早的评论

Peter Perkins 2024-8-23

This is a pretty advanced move. It's worth learning why it works, because there are several things going on that are useful to know:

RawFileIndex is a table with 36 rows and a bunch of variables
Its Parameters variable is a cell array. Tables can contain cell arrays
Each of the 36 cells in RawFileIndex.Parameters contains a table, all of them with the same two variables, but with different numbers of rows. So that cell array allows you to encapsulate multiple rows of parameters for each row of RawFileIndex
RawFileIndex.Parameters{:} expands those 36 cells into what's called a "comma-separated list", which is more or less a stream of 36 separate values, i..e. 36 tables
vertcat(RawFileIndex.Parameters{:}) captures that stream and vertically concatenates the 36 tables. They all have the same variables, so that works. The result is one Nx36 table, where N is something like 36*33
vertcat(RawFileIndex.Parameters{:}).ID extracts the ID variable of that Nx2 table. It's an Nx1 string column vector, because all of the ID variables in the 36 separate tables were string columns
Finally, call unique on that string column

Stephen23 2024-8-29

编辑：Stephen23 2024-8-29

"was it the right choice to use a table for it?"

If a table holds the information you require in a way that makes accessing it efficient and easy to understand for you, then it was a reasonable choice.

"whould you have structered it any differently?"

I probably would have used a structure array. In general use the simplest data structure that can reasonably contain the data.

"is it possible to have a table directly within another table or does it alway have to be a table in a cell within a table?"

Tables must be in a cell array.

Every table column actually consists of one array, which therefore must have exactly as many rows as the table has. However, in general tables have arbitrary number of rows, so in general the number of rows of nested tables would not sum to the number of rows of the parent table. Therefore each nested table is instead placed in a cell array (which does have exactly the same number of rows as the parent table). Exactly the same argument applies to all other arrays where non-scalar rows must be stored in one table row.

Paolo Mazzoleni 2024-8-29