filtering excel commands in matlab

6 Ansichten (letzte 30 Tage)
fima v
fima v am 10 Sep. 2022
Kommentiert: Star Strider am 11 Sep. 2022
Hello, i have the attched file .
i want using script command to extract the table and filter the rows of the table where the dogs are.
so it basicky out of my excel table i need to get 3D array of of 1X1X2 (animal,size,Color)
dog M brown
dog M red
is there an effective way to do it in matlab commands?
Thanks.

Antworten (2)

dpb
dpb am 10 Sep. 2022
Bearbeitet: dpb am 11 Sep. 2022
tA=readtable('https://www.mathworks.com/matlabcentral/answers/uploaded_files/1121415/animals.xlsx');
tA=convertvars(tA,@iscellstr,'categorical');
tA
tA = 6×3 table
Animal size color ________ ____ ______ cat s black cat ss orange cat sss white dog M brown elephant L grey dog MM red
Now we've got the table, don't make piecemeal of it, use the variables of interest as grouping variables instead...
groupsummary(tA,'Animal')
ans = 3×2 table
Animal GroupCount ________ __________ cat 3 dog 2 elephant 1
When you subsequently add other data such as weights, ages, etc., etc., ... then you can compute statistics or whatever the same way with whatever functions you need.
tA.Animal=='dog'% see the logical addressing vector
ans = 6×1 logical array
0 0 0 1 0 1
tD=tA(tA.Animal=='dog',{'size','color'})
tD = 2×2 table
size color ____ _____ M brown MM red
illustrates picking only the non-grouping variables since 'Animal' is now superfluous. To continue to carry it along anyway, use ":" for the column addressing vector.
  2 Kommentare
fima v
fima v am 11 Sep. 2022
Hello dpb, i am looking for getting a data structure which includes only the 'dog' rows form the table
" dog M brown"
"dog MM red"
what is the most effective method to get these two rows alone?
Thanks.
dpb
dpb am 11 Sep. 2022
In general, it's wrong approach to physically separate tables unless it really is known that aren't going to use any of the other data for further analysis, but if you're adamant --
tDog=tA(tA.Animal=='dog',:);

Melden Sie sich an, um zu kommentieren.


Star Strider
Star Strider am 10 Sep. 2022
I have no idea what you want or the reason a 3D matrix is necessary.
A different approach using unstack
T1 = readtable('https://www.mathworks.com/matlabcentral/answers/uploaded_files/1121415/animals.xlsx')
T1 = 6×3 table
Animal size color ____________ _______ __________ {'cat' } {'s' } {'black' } {'cat' } {'ss' } {'orange'} {'cat' } {'sss'} {'white' } {'dog' } {'M' } {'brown' } {'elephant'} {'L' } {'grey' } {'dog' } {'MM' } {'red' }
T1u = unstack(T1,'size','Animal')
T1u = 6×4 table
color cat dog elephant __________ __________ __________ __________ {'black' } {'s' } {0×0 char} {0×0 char} {'orange'} {'ss' } {0×0 char} {0×0 char} {'white' } {'sss' } {0×0 char} {0×0 char} {'brown' } {0×0 char} {'M' } {0×0 char} {'grey' } {0×0 char} {0×0 char} {'L' } {'red' } {0×0 char} {'MM' } {0×0 char}
Different results are available with different argument permutations and more arguments as described in Name-Value Arguments.
.
  4 Kommentare
fima v
fima v am 11 Sep. 2022
Bearbeitet: fima v am 11 Sep. 2022
Hello Star strider,you method works great,
could you please help me with a method of multiple criteria?
i want animal to be a "dog" size to be "MM" and color "red".
So the output will be only the row shown bellow.
{'dog'} {'MM'} {'red' }
is there some method to do it in my example?
Thanks.
Star Strider
Star Strider am 11 Sep. 2022
Try this —
T1 = readtable('https://www.mathworks.com/matlabcentral/answers/uploaded_files/1121415/animals.xlsx')
T1 = 6×3 table
Animal size color ____________ _______ __________ {'cat' } {'s' } {'black' } {'cat' } {'ss' } {'orange'} {'cat' } {'sss'} {'white' } {'dog' } {'M' } {'brown' } {'elephant'} {'L' } {'grey' } {'dog' } {'MM' } {'red' }
Dogs = T1(strcmpi(T1.Animal,'dog') & strcmpi(T1.color,'red'),:)
Dogs = 1×3 table
Animal size color _______ ______ _______ {'dog'} {'MM'} {'red'}
.

Melden Sie sich an, um zu kommentieren.

Community Treasure Hunt

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

Start Hunting!

Translated by