20.23 Complexity Parameter Table

We can look at the raw data to have a more precise and detailed view of the data. Here we only list specific rows from the complexity parameter table.

tmodel$cptable[c(1:5,22:29, 80:83),]
##              CP nsplit rel error    xerror        xstd
## 1  1.453731e-01      0 1.0000000 1.0000000 0.004616611
## 2  3.460366e-02      1 0.8546269 0.8546269 0.004349444
## 3  6.877579e-03      3 0.7854195 0.7894112 0.004214897
## 4  4.814305e-03      4 0.7785420 0.7856353 0.004206800
## 5  3.506217e-03      8 0.7592847 0.7633573 0.004158311
## 22 5.394180e-04     40 0.7199342 0.7314238 0.004086605
## 23 5.259325e-04     41 0.7193948 0.7308574 0.004085309
## 24 5.214374e-04     52 0.7130566 0.7302910 0.004084012
## 25 5.124471e-04     55 0.7114923 0.7302640 0.004083951
## 26 4.854762e-04     57 0.7104674 0.7300752 0.004083518
## 27 4.719907e-04     60 0.7090110 0.7294279 0.004082035
## 28 4.585053e-04     64 0.7071230 0.7294279 0.004082035
## 29 4.315344e-04     66 0.7062060 0.7285649 0.004080055
## 80 8.630687e-05    791 0.5906627 0.7232246 0.004067761
## 81 8.540784e-05    807 0.5888017 0.7235213 0.004068446
## 82 8.091270e-05    815 0.5880735 0.7241686 0.004069940
## 83 7.754133e-05   1012 0.5709469 0.7263533 0.004074973

%$

See how the relative error continues to decrease as the tree becomes more complex, but the cross validated error decreases and then starts to increase! We might choose a sensible value of from this table.



Your donation will support ongoing development and give you access to the PDF version of this book. Desktop Survival Guides include Data Science, GNU/Linux, and MLHub. Books available on Amazon include Data Mining with Rattle and Essentials of Data Science. Popular open source software includes rattle, wajig, and mlhub. Hosted by Togaware, a pioneer of free and open source software since 1984.
Copyright © 1995-2021 Graham.Williams@togaware.com Creative Commons Attribution-ShareAlike 4.0.