run multiple linear regression using fitlm without create new table?

3 次查看（过去 30 天）
Liqun
Liqun 2016-12-21
编辑： Liqun 2016-12-22
I have a large table (500000 rows X 26 columns), there are two category variables, say c1 with value +1, 0 and -1; c2 with value a, b, and c. I need to run 16 regressions using fitlm, which are listed in the table
ALL a b c
ALL
+1
0
-1
(ALL, ALL) means using all the data in the table;
(ALL, a) means using data c2 = a, c1 can be any;
(+1, a) means using data c1 = +1 and c2 =a;
so on and so forth.
Right now in my mind, I can pull out the sub-table 16 times from the large tabel, then run the regression.
My question: is there any more efficient way to do this, without creating 16 sub-tables?
Thanks!

请先登录，再回答此问题。

回答（1 个）

David Barry
David Barry 2016-12-21
编辑：David Barry 2016-12-21
If you need to operate on a subset of the data then ultimately you will need to create that subset of the data. You could try using a categorical array to reduce your memory footprint though.
https://uk.mathworks.com/help/matlab/ref/categorical.html https://uk.mathworks.com/help/matlab/matlab_prog/advantages-of-using-categorical-arrays.html

类别

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

标签

Community Treasure Hunt

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

Start Hunting!

Translated by