[R] rpart

carol white wht_crl at yahoo.com
Sun Jan 27 20:31:30 CET 2013


Hi,
When I look at the summary of an rpart object run on my data, I get 7 nodes but when I plot the rpart object, I get only 3 nodes. Should the number of nodes not match in the results of the 2 functions (summary and plot) or it is not always the same?

Look forward to your reply,

Carol
--------------------------------------------
 summary(rpart.res)
Call:
rpart(formula = mydata$class ~ ., data = as.data.frame(t(mydata)))
  n= 62 

         CP nsplit rel error    xerror      xstd
1 0.6363636      0 1.0000000 1.0000000 0.1712469
2 0.1363636      1 0.3636364 0.6818182 0.1532767
3 0.0100000      2 0.2272727 0.7727273 0.1596659

Variable importance
  Hsa.627   Hsa.692 Hsa.692.2  Hsa.3306   Hsa.601   Hsa.831  Hsa.1832  Hsa.2456 
       19        13        11        10        10         8         6         6 
 Hsa.8147  Hsa.1131 Hsa.692.1 
        6         5         5 

Node number 1: 62 observations,    complexity param=0.6363636
  predicted class=t  expected loss=0.3548387  P(node) =1
    class counts:    22    40
   probabilities: 0.355 0.645 
  left son=2 (14 obs) right son=3 (48 obs)
  Primary splits:
      Hsa.627   < 59.83    to the left,  improve=15.05376, (0 missing)
      Hsa.8147  < 1696.23  to the right, improve=14.46790, (0 missing)
      Hsa.37937 < 379.39   to the right, improve=13.75358, (0 missing)
      Hsa.692.2 < 842.305  to the right, improve=12.38710, (0 missing)
      Hsa.1832  < 735.805  to the right, improve=11.90495, (0 missing)
  Surrogate splits:
      Hsa.692.2 < 1086.655 to the right, agree=0.903, adj=0.571, (0 split)
      Hsa.3306  < 170.515  to the left,  agree=0.887, adj=0.500, (0 split)
      Hsa.601   < 88.065   to the left,  agree=0.887, adj=0.500, (0 split)
      Hsa.692   < 1251.99  to the right, agree=0.871, adj=0.429, (0 split)
      Hsa.831   < 281.54   to the left,  agree=0.871, adj=0.429, (0 split)

Node number 2: 14 observations
  predicted class=n  expected loss=0  P(node) =0.2258065
    class counts:    14     0
   probabilities: 1.000 0.000 

Node number 3: 48 observations,    complexity param=0.1363636
  predicted class=t  expected loss=0.1666667  P(node) =0.7741935
    class counts:     8    40
   probabilities: 0.167 0.833 
  left son=6 (7 obs) right son=7 (41 obs)
  Primary splits:
      Hsa.8147  < 1722.605 to the right, improve=4.915215, (0 missing)
      Hsa.1832  < 681.145  to the right, improve=4.915215, (0 missing)
      Hsa.1410  < 49.985   to the left,  improve=4.915215, (0 missing)
      Hsa.2456  < 186.195  to the right, improve=4.915215, (0 missing)
      Hsa.11616 < 969.085  to the right, improve=4.915215, (0 missing)
  Surrogate splits:
      Hsa.1832  < 681.145  to the right, agree=1.000, adj=1.000, (0 split)
      Hsa.2456  < 186.195  to the right, agree=1.000, adj=1.000, (0 split)
      Hsa.692   < 1048.375 to the right, agree=0.979, adj=0.857, (0 split)
      Hsa.692.1 < 1136.75  to the right, agree=0.979, adj=0.857, (0 split)
      Hsa.1131  < 1679.54  to the right, agree=0.979, adj=0.857, (0 split)

Node number 6: 7 observations
  predicted class=n  expected loss=0.2857143  P(node) =0.1129032
    class counts:     5     2
   probabilities: 0.714 0.286 

Node number 7: 41 observations
  predicted class=t  expected loss=0.07317073  P(node) =0.6612903
    class counts:     3    38
   probabilities: 0.073 0.927 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: rpart.png
Type: image/png
Size: 12105 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20130127/849451f7/attachment.png>


More information about the R-help mailing list