Stepwiseglm with non-binary response variable (classifications into 3 buckets) and mix of continous/categorical predictor variables

1 次查看（过去 30 天）
Dhruv Ghulati
Dhruv Ghulati 2015-12-24
Hi there,
I have an array of 25 predictors that looks like this (not, converted from values like "Albania" to numbers like 46 for each country, for example, for categorical variables, and some with continuous variables):
0 5 841 93 10679 1101 1 161 50 3 3 1 1
0 0 0 0 0 0 14 161 50 13 9 1 1
0 0 0 0 0 0 3 161 83 13 3 1 1
0 0 5 0 17 41 22 161 83 3 2 1 1
0 0 0 0 0 0 4 161 83 3 3 1 1
0 0 0 0 0 0 8 161 71 3 3 1 1
6 4 12 3 0 0 4 161 83 3 3 1 2
0 7 1 0 0 0 13 161 71 3 3 1 2
0 1 262 23 961 96 102 55 198 5 3 1 1
0 5 285 23 968 96 12 161 83 3 3 1 1
0 0 843 95 10681 1101 18 71 116 13 3 1 1
0 0 0 1 0 0 1 104 78 5 3 2 1
and an array of 1 response variables that looks like (note, converted from text to being ordinal variable):
1
2
3
2
1
3
I ran
stepwiseglm(predictormatrix,response);
And got the below output:
1. Adding x16, Deviance = 1614.387, FStat = 1743.1408, PValue = 0
2. Adding x25, Deviance = 1440.549, FStat = 2142.3401, PValue = 0
3. Adding x19, Deviance = 1363.947, FStat = 996.9882, PValue = 6.0490784e-213
4. Adding x2, Deviance = 1318.3645, FStat = 613.74033, PValue = 3.1244367e-133
5. Adding x6, Deviance = 1302.0096, FStat = 222.96267, PValue = 4.115593e-50
6. Adding x4, Deviance = 1286.8418, FStat = 209.20522, PValue = 3.7970611e-47
7. Adding x13, Deviance = 1270.84, FStat = 223.47373, PValue = 3.1943538e-50
8. Adding x7, Deviance = 1260.2216, FStat = 149.5337, PValue = 3.0111563e-34
9. Adding x21, Deviance = 1252.7396, FStat = 105.98862, PValue = 8.7078938e-25
10. Adding x4:x21, Deviance = 1244.1865, FStat = 121.98614, PValue = 2.8737974e-28
11. Adding x16:x21, Deviance = 1237.2232, FStat = 99.866125, PValue = 1.8808007e-23
12. Adding x2:x4, Deviance = 1231.7667, FStat = 78.598656, PValue = 8.3183569e-19
13. Adding x3, Deviance = 1225.9525, FStat = 84.143346, PValue = 5.094845e-20
14. Adding x3:x21, Deviance = 1221.1573, FStat = 69.665335, PValue = 7.5377414e-17
15. Adding x6:x13, Deviance = 1217.5627, FStat = 52.372501, PValue = 4.7786344e-13
16. Adding x2:x7, Deviance = 1215.1946, FStat = 34.569465, PValue = 4.1868738e-09
17. Adding x19:x21, Deviance = 1213.154, FStat = 29.835612, PValue = 4.7659952e-08
....
How should I interpret this? Is this telling me that I should choose the combo that was output with the lowest p value, and highest F stat? Am I missing certain modelspec and name value pair arguments within this link that would make much more sense given my data?
Thanks,

请先登录，再回答此问题。

回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Gaussian Process Regression 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