Regression and Classification Trees
library(rwf)Description
This shows the output of PART RPART functions from rwf.
Installation instructions for rwf can be found here These functions are not
installed by default they are in
/rwf/working_functions/OTHER/ML_TREE_RPART.R
infert_formula<-formula(factor(case)~age+parity+education+spontaneous+induced)
boston_formula<-formula(c("medv~",paste(names(MASS::Boston)[1:13],collapse="+")))## Warning in formula.character(c("medv~", paste(names(MASS::Boston)[1:13], : Using formula(x) is deprecated when x is a character vector of length > 1.
## Consider formula(paste(x, collapse = " ")) instead.print(infert_formula)## factor(case) ~ age + parity + education + spontaneous + induced
print(boston_formula)## medv ~ crim + zn + indus + chas + nox + rm + age + dis + rad + ## tax + ptratio + black + lstat
# kfolding
train_test_classification<-k_fold(df=infert,
model_formula=infert_formula)## Fold Cases: 1 Train: 223 Test: 25 Total: 248 Unique Train: 223 Unique Test: 25 ## Fold Cases: 2 Train: 223 Test: 25 Total: 248 Unique Train: 223 Unique Test: 25 ## Fold Cases: 3 Train: 223 Test: 25 Total: 248 Unique Train: 223 Unique Test: 25 ## Fold Cases: 4 Train: 224 Test: 24 Total: 248 Unique Train: 224 Unique Test: 24 ## Fold Cases: 5 Train: 223 Test: 25 Total: 248 Unique Train: 223 Unique Test: 25 ## Fold Cases: 6 Train: 223 Test: 25 Total: 248 Unique Train: 223 Unique Test: 25 ## Fold Cases: 7 Train: 224 Test: 24 Total: 248 Unique Train: 224 Unique Test: 24 ## Fold Cases: 8 Train: 223 Test: 25 Total: 248 Unique Train: 223 Unique Test: 25 ## Fold Cases: 9 Train: 223 Test: 25 Total: 248 Unique Train: 223 Unique Test: 25 ## Fold Cases: 10 Train: 223 Test: 25 Total: 248 Unique Train: 223 Unique Test: 25
train_test_regression<-k_fold(df=MASS::Boston,
model_formula=boston_formula)## Fold Cases: 1 Train: 455 Test: 51 Total: 506 Unique Train: 455 Unique Test: 51 ## Fold Cases: 2 Train: 455 Test: 51 Total: 506 Unique Train: 455 Unique Test: 51 ## Fold Cases: 3 Train: 456 Test: 50 Total: 506 Unique Train: 456 Unique Test: 50 ## Fold Cases: 4 Train: 455 Test: 51 Total: 506 Unique Train: 455 Unique Test: 51 ## Fold Cases: 5 Train: 456 Test: 50 Total: 506 Unique Train: 456 Unique Test: 50 ## Fold Cases: 6 Train: 455 Test: 51 Total: 506 Unique Train: 455 Unique Test: 51 ## Fold Cases: 7 Train: 456 Test: 50 Total: 506 Unique Train: 456 Unique Test: 50 ## Fold Cases: 8 Train: 455 Test: 51 Total: 506 Unique Train: 455 Unique Test: 51 ## Fold Cases: 9 Train: 456 Test: 50 Total: 506 Unique Train: 456 Unique Test: 50 ## Fold Cases: 10 Train: 455 Test: 51 Total: 506 Unique Train: 455 Unique Test: 51
# srpart models
rtree_classification<-rpart::rpart(infert_formula,
train_test_classification$f$train$f1,
model=TRUE,x=TRUE,y=TRUE)
rtree_regression<-rpart::rpart(boston_formula,
train_test_regression$f$train$f1,
model=TRUE,x=TRUE,y=TRUE)Rpart Classification
result<-data.frame(rtree_classification$cptable)
result$nsplit<-factor(result$nsplit+1)
minimun_size<-as.numeric(as.character(result[which.min(result[,"xerror"]),"nsplit"]))
initial_model<-rtree_classification
model<-rpart::prune(rtree_classification,
cp=rtree_classification$cptable[which.min(rtree_classification$cptable[,"xerror"]),"CP"])
importance<-model$variable.importance
importance<-data.frame(names=names(importance),importance=importance)
importance$names<-factor(importance$names,levels=rev(as.character(importance$names)))
plot_importance<-ggplot(importance,aes(x=names,y=importance))+
geom_bar(stat='identity')+
labs(title="Importance Plot",y="Relative Influence",x="Predictor")+
theme_bw(base_size=10)+
scale_x_discrete(limits=rev(levels(names)))+
coord_flip()
plot_importance
rtree_classification## n= 223 ## ## node), split, n, loss, yval, (yprob) ## * denotes terminal node ## ## 1) root 223 74 0 (0.6681614 0.3318386) ## 2) spontaneous< 0.5 128 25 0 (0.8046875 0.1953125) * ## 3) spontaneous>=0.5 95 46 1 (0.4842105 0.5157895) ## 6) age< 30.5 52 20 0 (0.6153846 0.3846154) ## 12) parity>=2.5 19 4 0 (0.7894737 0.2105263) * ## 13) parity< 2.5 33 16 0 (0.5151515 0.4848485) ## 26) spontaneous< 1.5 26 10 0 (0.6153846 0.3846154) ## 52) parity>=1.5 9 1 0 (0.8888889 0.1111111) * ## 53) parity< 1.5 17 8 1 (0.4705882 0.5294118) * ## 27) spontaneous>=1.5 7 1 1 (0.1428571 0.8571429) * ## 7) age>=30.5 43 14 1 (0.3255814 0.6744186) ## 14) parity>=3.5 10 4 0 (0.6000000 0.4000000) * ## 15) parity< 3.5 33 8 1 (0.2424242 0.7575758) *
rpart::plotcp(model)
rpart::rsq.rpart(model)## ## Classification tree: ## rpart::rpart(formula = infert_formula, data = train_test_classification$f$train$f1, ## model = TRUE, x = TRUE, y = TRUE) ## ## Variables actually used in tree construction: ## [1] age spontaneous ## ## Root node error: 74/223 = 0.33184 ## ## n= 223 ## ## CP nsplit rel error xerror xstd ## 1 0.101351 0 1.0000 1.00000 0.095022 ## 2 0.033784 2 0.7973 0.83784 0.090412
## Warning in rpart::rsq.rpart(model): may not be applicable for this method
error<-data.frame(model$cptable)
error$nsplit<-factor(error$nsplit+1)
tree_size<-error[which.min(error[,"xerror"]),"nsplit"]
error<-reshape2::melt(error,id.vars="nsplit")
names(error)<-c("Split","Metric","value")
plot_prune<-ggplot(error,aes(x=Split,y=value,color=Metric))+
geom_line(aes(group=Metric))+
geom_point()+
labs(title=paste("Error Plot","Suggested Size:",tree_size),y="Metric value",x="Size of Tree")+
theme_bw(base_size=10)
plot_prune
rpart.plot::rpart.plot(model,type=1)
frame<-data.frame(model$frame)
cp<-data.frame(model$cptable)
parameters<-data.frame(parameters=unlist(model$control))
splits<-data.frame(name=row.names(model$splits),model$splits,row.names=NULL)
importance<-data.frame(importance=model$variable.importance)
ordered<-data.frame(ordered=model$ordered)
data<-data.frame(y=model$y,x=model$x,model=model$model)
call<-data.frame(call=call_to_string(model))
result<-list(frame=frame,cp=cp,parameters=parameters,splits=splits,importance=importance,ordered=ordered,call=call)
print(result)## $frame ## var n wt dev yval complexity ncompete nsurrogate yval2.V1 yval2.V2 yval2.V3 yval2.V4 yval2.V5 yval2.nodeprob ## 1 spontaneous 223 223 74 1 0.10135135 4 3 1.0000000 149.0000000 74.0000000 0.6681614 0.3318386 1.0000000 ## 2128 128 25 1 0.00000000 0 0 1.0000000 103.0000000 25.0000000 0.8046875 0.1953125 0.5739910 ## 3 age 95 95 46 2 0.10135135 4 3 2.0000000 46.0000000 49.0000000 0.4842105 0.5157895 0.4260090 ## 6 52 52 20 1 0.03378378 0 0 1.0000000 32.0000000 20.0000000 0.6153846 0.3846154 0.2331839 ## 7 43 43 14 2 0.02702703 0 0 2.0000000 14.0000000 29.0000000 0.3255814 0.6744186 0.1928251 ## ## $cp ## CP nsplit rel.error xerror xstd ## 1 0.10135135 0 1.0000000 1.0000000 0.09502215 ## 2 0.03378378 2 0.7972973 0.8378378 0.09041170 ## ## $parameters ## parameters ## minsplit 20.00 ## minbucket 7.00 ## cp 0.01 ## maxcompete 4.00 ## maxsurrogate 5.00 ## usesurrogate 2.00 ## surrogatestyle 0.00 ## maxdepth 30.00 ## xval 10.00 ## ## $splits ## name count ncat improve index adj ## 1 spontaneous 223 -1 11.200885798 0.5 0.00000000 ## 2 age 223 -1 0.323644455 25.5 0.00000000 ## 3 education 223 3 0.080859787 1.0 0.00000000 ## 4 parity 223 -1 0.076903366 4.5 0.00000000 ## 5 induced 223 1 0.027944733 1.5 0.00000000 ## 6 parity 0 -1 0.636771300 2.5 0.14736842 ## 7 induced 0 1 0.618834081 0.5 0.10526316 ## 8 age 0 1 0.578475336 22.0 0.01052632 ## 9 age 95 -1 3.953526033 30.5 0.00000000 ## 10 spontaneous 95 -1 2.845488722 1.5 0.00000000 ## 11 parity 95 1 2.288605484 2.5 0.00000000 ## 12 induced 95 1 0.212021250 0.5 0.00000000 ## 13 education 95 3 0.007890046 2.0 0.00000000 ## 14 spontaneous 0 -1 0.589473684 1.5 0.09302326 ## 15 parity 0 -1 0.557894737 5.5 0.02325581 ## 16 education 0 3 0.557894737 3.0 0.02325581 ## ## $importance ## importance ## spontaneous 11.56865566 ## age 4.07143009 ## parity 1.74259932 ## induced 1.17904061 ## education 0.09194247 ## ## $ordered ## ordered ## age FALSE ## parity FALSE ## education FALSE ## spontaneous FALSE ## induced FALSE ## ## $call ## call ## 1 rpart::rpart(formula=infert_formula,data=train_test_classification$f$train$f1,,model=TRUE,x=TRUE,y=TRUE)
Rpart Regression
result<-data.frame(rtree_regression$cptable)
result$nsplit<-factor(result$nsplit+1)
minimun_size<-as.numeric(as.character(result[which.min(result[,"xerror"]),"nsplit"]))
initial_model<-rtree_regression
model<-rpart::prune(rtree_regression,cp=rtree_regression$cptable[which.min(rtree_regression$cptable[,"xerror"]),"CP"])
importance<-model$variable.importance
importance<-data.frame(names=names(importance),importance=importance)
importance$names<-factor(importance$names,levels=rev(as.character(importance$names)))
plot_importance<-ggplot(importance,aes(x=names,y=importance))+
geom_bar(stat='identity')+
labs(title="Importance Plot",y="Relative Influence",x="Predictor")+
theme_bw(base_size=10)+
scale_x_discrete(limits=rev(levels(names)))+
coord_flip()
plot_importance
rtree_regression## n= 455 ## ## node), split, n, deviance, yval ## * denotes terminal node ## ## 1) root 455 38518.7300 22.79846 ## 2) rm< 6.941 385 15239.2800 20.12571 ## 4) lstat>=14.4 154 2926.3910 15.23506 ## 8) nox>=0.603 95 1238.1250 13.23158 ## 16) lstat>=19.645 46 401.5122 10.81304 * ## 17) lstat< 19.645 49 314.9498 15.50204 * ## 9) nox< 0.603 59 692.9403 18.46102 * ## 5) lstat< 14.4 231 6173.8160 23.38615 ## 10) dis>=1.5511 224 3203.1260 22.92946 ## 20) rm< 6.5255 171 1402.6330 21.63216 * ## 21) rm>=6.5255 53 584.1679 27.11509 * ## 11) dis< 1.5511 7 1429.0200 38.00000 * ## 3) rm>=6.941 70 5402.6500 37.49857 ## 6) rm< 7.437 43 1829.7580 32.26512 ## 12) lstat>=9.65 7 432.9971 23.05714 * ## 13) lstat< 9.65 36 687.8489 34.05556 * ## 7) rm>=7.437 27 519.5200 45.83333 *
rpart::plotcp(model)
rpart::rsq.rpart(model)## ## Regression tree: ## rpart::rpart(formula = boston_formula, data = train_test_regression$f$train$f1, ## model = TRUE, x = TRUE, y = TRUE) ## ## Variables actually used in tree construction: ## [1] dis lstat nox rm ## ## Root node error: 38519/455 = 84.657 ## ## n= 455 ## ## CP nsplit rel error xerror xstd ## 1 0.464107 0 1.00000 1.00634 0.087379 ## 2 0.159379 1 0.53589 0.59057 0.060113 ## 3 0.079270 2 0.37651 0.38982 0.047756 ## 4 0.040024 3 0.29724 0.32012 0.042891 ## 5 0.031577 4 0.25722 0.32899 0.045074 ## 6 0.025840 5 0.22564 0.30701 0.044923 ## 7 0.018404 6 0.19980 0.27651 0.043060 ## 8 0.013543 7 0.18140 0.27456 0.042179 ## 9 0.010000 8 0.16786 0.25973 0.038528
error<-data.frame(model$cptable)
error$nsplit<-factor(error$nsplit+1)
tree_size<-error[which.min(error[,"xerror"]),"nsplit"]
error<-reshape2::melt(error,id.vars="nsplit")
names(error)<-c("Split","Metric","value")
plot_prune<-ggplot(error,aes(x=Split,y=value,color=Metric))+
geom_line(aes(group=Metric))+
geom_point()+
labs(title=paste("Error Plot","Suggested Size:",tree_size),y="Metric value",x="Size of Tree")+
theme_bw(base_size=10)
plot_prune
rpart.plot::rpart.plot(model,type=1)
frame<-data.frame(model$frame)
cp<-data.frame(model$cptable)
parameters<-data.frame(parameters=unlist(model$control))
splits<-data.frame(name=row.names(model$splits),model$splits,row.names=NULL)
importance<-data.frame(importance=model$variable.importance)
ordered<-data.frame(ordered=model$ordered)
data<-data.frame(y=model$y,x=model$x,model=model$model)
call<-data.frame(call=call_to_string(model))
result<-list(frame=frame,cp=cp,parameters=parameters,splits=splits,importance=importance,ordered=ordered,call=call)
print(result)## $frame ## var n wt dev yval complexity ncompete nsurrogate ## 1 rm 455 455 38518.7289 22.79846 0.464106790 4 5 ## 2 lstat 385 385 15239.2754 20.12571 0.159378808 4 5 ## 4 nox 154 154 2926.3906 15.23506 0.025840028 4 5 ## 8 lstat 95 95 1238.1253 13.23158 0.013543108 4 5 ## 1646 46 401.5122 10.81304 0.004020501 0 0 ## 17 49 49 314.9498 15.50204 0.010000000 0 0 ## 9 59 59 692.9403 18.46102 0.004452308 0 0 ## 5 dis 231 231 6173.8157 23.38615 0.040023910 4 1 ## 10 rm 224 224 3203.1255 22.92946 0.031577483 4 5 ## 20 171 171 1402.6331 21.63216 0.006274422 0 0 ## 21 53 53 584.1679 27.11509 0.004440859 0 0 ## 11 7 7 1429.0200 38.00000 0.010000000 0 0 ## 3 rm 70 70 5402.6499 37.49857 0.079269806 4 5 ## 6 lstat 43 43 1829.7577 32.26512 0.018404336 4 5 ## 12 7 7 432.9971 23.05714 0.010000000 0 0 ## 13 36 36 687.8489 34.05556 0.005420060 0 0 ## 7 27 27 519.5200 45.83333 0.004164039 0 0 ## ## $cp ## CP nsplit rel.error xerror xstd ## 1 0.46410679 0 1.0000000 1.0063399 0.08737869 ## 2 0.15937881 1 0.5358932 0.5905702 0.06011321 ## 3 0.07926981 2 0.3765144 0.3898214 0.04775561 ## 4 0.04002391 3 0.2972446 0.3201161 0.04289143 ## 5 0.03157748 4 0.2572207 0.3289940 0.04507448 ## 6 0.02584003 5 0.2256432 0.3070061 0.04492335 ## 7 0.01840434 6 0.1998032 0.2765072 0.04305981 ## 8 0.01354311 7 0.1813988 0.2745644 0.04217893 ## 9 0.01000000 8 0.1678557 0.2597300 0.03852842 ## ## $parameters ## parameters ## minsplit 20.00 ## minbucket 7.00 ## cp 0.01 ## maxcompete 4.00 ## maxsurrogate 5.00 ## usesurrogate 2.00 ## surrogatestyle 0.00 ## maxdepth 30.00 ## xval 10.00 ## ## $splits ## name count ncat improve index adj ## 1 rm 455 -1 0.46410679 6.941000 0.00000000 ## 2 lstat 455 1 0.44229881 9.725000 0.00000000 ## 3 indus 455 1 0.25873730 6.660000 0.00000000 ## 4 ptratio 455 1 0.23246618 19.650000 0.00000000 ## 5 nox 455 1 0.21323307 0.669500 0.00000000 ## 6 lstat 0 1 0.89230769 4.830000 0.30000000 ## 7 ptratio 0 1 0.87252747 14.550000 0.17142857 ## 8 zn 0 -1 0.85714286 87.500000 0.07142857 ## 9 indus 0 1 0.85494505 1.605000 0.05714286 ## 10 crim 0 1 0.84835165 0.013355 0.01428571 ## 11 lstat 385 1 0.40284521 14.400000 0.00000000 ## 12 nox 385 1 0.26198079 0.669500 0.00000000 ## 13 crim 385 1 0.22568317 5.848030 0.00000000 ## 14 ptratio 385 1 0.19631370 19.900000 0.00000000 ## 15 age 385 1 0.19083383 75.750000 0.00000000 ## 16 age 0 1 0.82077922 88.100000 0.55194805 ## 17 indus 0 1 0.77922078 16.570000 0.44805195 ## 18 nox 0 1 0.77922078 0.576500 0.44805195 ## 19 dis 0 -1 0.77922078 2.239350 0.44805195 ## 20 tax 0 1 0.77142857 434.500000 0.42857143 ## 21 nox 154 1 0.34012036 0.603000 0.00000000 ## 22 crim 154 1 0.33597887 7.464950 0.00000000 ## 23 tax 154 1 0.28580671 567.500000 0.00000000 ## 24 dis 154 -1 0.28185524 1.986400 0.00000000 ## 25 ptratio 154 1 0.25070026 19.450000 0.00000000 ## 26 indus 0 1 0.88311688 15.995000 0.69491525 ## 27 tax 0 1 0.88311688 397.000000 0.69491525 ## 28 dis 0 -1 0.83766234 2.790850 0.57627119 ## 29 crim 0 1 0.82467532 1.400920 0.54237288 ## 30 ptratio 0 1 0.75974026 19.900000 0.37288136 ## 31 lstat 95 1 0.42133321 19.645000 0.00000000 ## 32 crim 95 1 0.40969185 11.343000 0.00000000 ## 33 dis 95 -1 0.19481532 2.003700 0.00000000 ## 34 tax 95 1 0.15909053 551.500000 0.00000000 ## 35 rm 95 -1 0.13723275 5.453500 0.00000000 ## 36 dis 0 -1 0.82105263 1.663450 0.63043478 ## 37 rm 0 -1 0.77894737 5.627500 0.54347826 ## 38 crim 0 1 0.73684211 10.533600 0.45652174 ## 39 age 0 1 0.69473684 98.850000 0.36956522 ## 40 nox 0 -1 0.66315789 0.706500 0.30434783 ## 41 dis 231 1 0.24971107 1.551100 0.00000000 ## 42 lstat 231 1 0.20793193 9.545000 0.00000000 ## 43 rm 231 -1 0.19764412 6.542000 0.00000000 ## 44 chas 231 -1 0.09958911 0.500000 0.00000000 ## 45 crim 231 -1 0.08938980 4.866945 0.00000000 ## 46 crim 0 -1 0.98268398 8.053285 0.42857143 ## 47 rm 224 -1 0.37973052 6.525500 0.00000000 ## 48 lstat 224 1 0.31435902 7.685000 0.00000000 ## 49 nox 224 1 0.14357212 0.512500 0.00000000 ## 50 tax 224 1 0.14323414 222.500000 0.00000000 ## 51 indus 224 1 0.14267172 4.220000 0.00000000 ## 52 lstat 0 1 0.81696429 5.055000 0.22641509 ## 53 crim 0 1 0.78125000 0.018370 0.07547170 ## 54 zn 0 -1 0.78125000 31.500000 0.07547170 ## 55 chas 0 -1 0.76785714 0.500000 0.01886792 ## 56 dis 0 -1 0.76785714 10.648000 0.01886792 ## 57 rm 70 -1 0.56516196 7.437000 0.00000000 ## 58 lstat 70 1 0.31465530 4.680000 0.00000000 ## 59 ptratio 70 1 0.17505508 19.150000 0.00000000 ## 60 crim 70 1 0.12219048 1.921980 0.00000000 ## 61 black 70 1 0.12167649 395.590000 0.00000000 ## 62 lstat 0 1 0.78571429 4.505000 0.44444444 ## 63 ptratio 0 1 0.65714286 14.750000 0.11111111 ## 64 black 0 1 0.65714286 389.885000 0.11111111 ## 65 crim 0 1 0.64285714 0.024530 0.07407407 ## 66 zn 0 -1 0.64285714 81.250000 0.07407407 ## 67 lstat 43 1 0.38743471 9.650000 0.00000000 ## 68 rad 43 1 0.20118078 7.500000 0.00000000 ## 69 nox 43 1 0.18477792 0.639000 0.00000000 ## 70 ptratio 43 1 0.15678520 18.900000 0.00000000 ## 71 indus 43 1 0.12461986 12.585000 0.00000000 ## 72 crim 0 1 0.90697674 0.724605 0.42857143 ## 73 nox 0 1 0.90697674 0.659000 0.42857143 ## 74 rad 0 1 0.88372093 16.000000 0.28571429 ## 75 tax 0 1 0.88372093 534.500000 0.28571429 ## 76 ptratio 0 1 0.88372093 19.700000 0.28571429 ## ## $importance ## importance ## rm 22430.01299 ## lstat 14365.13367 ## dis 5217.69336 ## indus 4463.82007 ## nox 4208.53329 ## ptratio 3977.54282 ## age 3581.23584 ## tax 3525.24236 ## crim 2315.87970 ## zn 1594.88834 ## black 339.26358 ## rad 202.54618 ## chas 22.94952 ## ## $ordered ## ordered ## crim FALSE ## zn FALSE ## indus FALSE ## chas FALSE ## nox FALSE ## rm FALSE ## age FALSE ## dis FALSE ## rad FALSE ## tax FALSE ## ptratio FALSE ## black FALSE ## lstat FALSE ## ## $call ## call ## 1 rpart::rpart(formula=boston_formula,data=train_test_regression$f$train$f1,,model=TRUE,x=TRUE,y=TRUE)