Paper - Recursive Partitioning

*The author of this computation has been verified*

R Software Module: /rwasp_regression_trees1.wasp (opens new window with default values)

Title produced by software: Recursive Partitioning (Regression Trees)

Date of computation: Tue, 21 Dec 2010 12:58:36 +0000

Cite this page as follows:

Statistical Computations at FreeStatistics.org, Office for Research Development and Education, URL http://www.freestatistics.org/blog/date/2010/Dec/21/t1292936273vvwkquavbdeta6c.htm/, Retrieved Tue, 21 Dec 2010 13:57:53 +0100

BibTeX entries for LaTeX users:

@Manual{KEY, author = {{YOUR NAME}}, publisher = {Office for Research Development and Education}, title = {Statistical Computations at FreeStatistics.org, URL http://www.freestatistics.org/blog/date/2010/Dec/21/t1292936273vvwkquavbdeta6c.htm/}, year = {2010}, } @Manual{R, title = {R: A Language and Environment for Statistical Computing}, author = {{R Development Core Team}}, organization = {R Foundation for Statistical Computing}, address = {Vienna, Austria}, year = {2010}, note = {{ISBN} 3-900051-07-0}, url = {http://www.R-project.org}, }

Original text written by user:

IsPrivate?

No (this computation is public)

User-defined keywords:

Dataseries X:

» Textbox « » Textfile « » CSV «

105.31 1576.23 29.29 710.45 105.63 1546.37 28.99 720 106.02 1545.05 28.91 720 105.85 1552.34 29.29 720 106.57 1594.3 30.96 754.78 106.48 1605.78 30.57 802.73 106.60 1673.21 30.59 845.24 106.75 1612.94 31.39 893.91 106.69 1566.34 31.28 931.43 106.69 1530.17 31.1 940 106.93 1582.54 31.7 947.73 107.21 1702.16 32.57 960 107.88 1701.93 32.49 996.96 108.84 1811.15 32.46 1000 108.96 1924.2 32.3 1000 109.52 2034.25 32.97 1000 108.45 2011.13 32.9 1013.04 108.67 2013.04 32.93 1095.24 108.96 2151.67 33.72 1159.09 108.76 1902.09 33.33 1200 107.85 1944.01 33.44 1200 108.78 1916.67 33.89 1282.61 107.51 1967.31 34.34 1513.64 108.83 2119.88 33.56 1669.05 111.54 2216.38 32.67 1700 111.74 2522.83 32.57 1700 112.04 2647.64 33.23 1700 111.74 2631.23 32.85 1665.91 111.81 2693.41 32.61 1650 111.86 3021.76 32.57 1650 114.23 2953.67 32.98 1619.57 114.80 2796.8 31.33 1599.05 115.17 2672.05 29.8 1572.73 115.11 2251.23 28.06 1470 114.43 2046.08 25.47 1268 114.66 2420.04 24.65 1217.39 115.11 2608.89 2 etc...

Output produced by software:

Enter (or paste) a matrix (table) containing all data (time) series. Every column represents a different variable and must be delimited by a space or Tab. Every row represents a period in time (or category) and must be delimited by hard returns. The easiest way to enter data is to copy and paste a block of spreadsheet cells. Please, do not use commas or spaces to seperate groups of digits! Summary of computational transaction Raw Input view raw input (R code) Raw Output view raw output of R engine Computing time 22 seconds R Server 'Sir Ronald Aylmer Fisher' @ 193.190.124.24 Goodness of Fit Correlation 0.9004 R-squared 0.8108 RMSE 2.5731 Actuals, Predictions, and Residuals # Actuals Forecasts Residuals 1 105.31 106.508461538462 -1.19846153846153 2 105.63 106.508461538462 -0.878461538461536 3 106.02 106.508461538462 -0.488461538461536 4 105.85 106.508461538462 -0.658461538461538 5 106.57 106.508461538462 0.0615384615384613 6 106.48 106.508461538462 -0.0284615384615279 7 106.6 106.508461538462 0.0915384615384625 8 106.75 106.508461538462 0.241538461538468 9 106.69 106.508461538462 0.181538461538466 10 106.69 106.508461538462 0.181538461538466 11 106.93 106.508461538462 0.421538461538475 12 107.21 106.508461538462 0.701538461538462 13 107.88 106.508461538462 1.37153846153846 14 108.84 109.315384615385 -0.475384615384613 15 108.96 109.315384615385 -0.355384615384622 16 109.52 109.315384615385 0.20461538461538 17 108.45 109.315384615385 -0.865384615384613 18 108.67 109.315384615385 -0.645384615384614 19 108.96 109.315384615385 -0.355384615384622 20 108.76 109.315384615385 -0.555384615384611 21 107.85 109.315384615385 -1.46538461538462 22 108.78 109.315384615385 -0.535384615384615 23 107.51 109.315384615385 -1.80538461538461 24 108.83 109.315384615385 -0.485384615384618 25 111.54 109.315384615385 2.22461538461539 26 111.74 113.166666666667 -1.42666666666668 27 112.04 113.166666666667 -1.12666666666667 28 111.74 113.166666666667 -1.42666666666668 29 111.81 113.166666666667 -1.35666666666667 30 111.86 113.166666666667 -1.30666666666667 31 114.23 113.166666666667 1.06333333333333 32 114.8 113.166666666667 1.63333333333333 33 115.17 113.166666666667 2.00333333333333 34 115.11 113.166666666667 1.94333333333333 35 114.43 109.315384615385 5.11461538461539 36 114.66 119.374090909091 -4.71409090909091 37 115.11 119.374090909091 -4.26409090909091 38 117.74 119.374090909091 -1.63409090909092 39 118.18 119.374090909091 -1.19409090909090 40 118.56 119.374090909091 -0.814090909090908 41 117.63 119.374090909091 -1.74409090909091 42 117.71 119.374090909091 -1.66409090909092 43 117.46 119.374090909091 -1.91409090909092 44 117.37 119.374090909091 -2.00409090909091 45 117.34 119.374090909091 -2.03409090909091 46 117.09 119.374090909091 -2.28409090909091 47 116.65 119.374090909091 -2.72409090909090 48 116.71 119.374090909091 -2.66409090909092 49 116.82 119.374090909091 -2.55409090909092 50 117.33 119.374090909091 -2.04409090909091 51 117.95 119.374090909091 -1.42409090909091 52 123.53 119.374090909091 4.15590909090909 53 124.91 119.374090909091 5.53590909090909 54 125.99 119.374090909091 6.61590909090908 55 126.29 119.374090909091 6.9159090909091 56 125.68 119.374090909091 6.3059090909091 57 125.52 119.374090909091 6.14590909090909

Charts produced by software:

http://www.freestatistics.org/blog/date/2010/Dec/21/t1292936273vvwkquavbdeta6c/2yg7f1292936292.png (open in new window) http://www.freestatistics.org/blog/date/2010/Dec/21/t1292936273vvwkquavbdeta6c/2yg7f1292936292.ps (open in new window) http://www.freestatistics.org/blog/date/2010/Dec/21/t1292936273vvwkquavbdeta6c/39p601292936292.png (open in new window) http://www.freestatistics.org/blog/date/2010/Dec/21/t1292936273vvwkquavbdeta6c/39p601292936292.ps (open in new window) http://www.freestatistics.org/blog/date/2010/Dec/21/t1292936273vvwkquavbdeta6c/4kgol1292936292.png (open in new window) http://www.freestatistics.org/blog/date/2010/Dec/21/t1292936273vvwkquavbdeta6c/4kgol1292936292.ps (open in new window)

Parameters (Session):

par1 = 1 ; par2 = none ; par4 = no ;

Parameters (R input):

par1 = 1 ; par2 = none ; par4 = no ;

R code (references can be found in the software module):

library(party) library(Hmisc) par1 <- as.numeric(par1) par3 <- as.numeric(par3) x <- data.frame(t(y)) is.data.frame(x) x <- x[!is.na(x[,par1]),] k <- length(x[1,]) n <- length(x[,1]) colnames(x)[par1] x[,par1] if (par2 == 'kmeans') { cl <- kmeans(x[,par1], par3) print(cl) clm <- matrix(cbind(cl$centers,1:par3),ncol=2) clm <- clm[sort.list(clm[,1]),] for (i in 1:par3) { cl$cluster[cl$cluster==clm[i,2]] <- paste('C',i,sep='') } cl$cluster <- as.factor(cl$cluster) print(cl$cluster) x[,par1] <- cl$cluster } if (par2 == 'quantiles') { x[,par1] <- cut2(x[,par1],g=par3) } if (par2 == 'hclust') { hc <- hclust(dist(x[,par1])^2, 'cen') print(hc) memb <- cutree(hc, k = par3) dum <- c(mean(x[memb==1,par1])) for (i in 2:par3) { dum <- c(dum, mean(x[memb==i,par1])) } hcm <- matrix(cbind(dum,1:par3),ncol=2) hcm <- hcm[sort.list(hcm[,1]),] for (i in 1:par3) { memb[memb==hcm[i,2]] <- paste('C',i,sep='') } memb <- as.factor(memb) print(memb) x[,par1] <- memb } if (par2=='equal') { ed <- cut(as.numeric(x[,par1]),par3,labels=paste('C',1:par3,sep='')) x[,par1] <- as.factor(ed) } table(x[,par1]) colnames(x) colnames(x)[par1] x[,par1] if (par2 == 'none') { m <- ctree(as.formula(paste(colnames(x)[par1],' ~ .',sep='')),data = x) } load(file='createtable') if (par2 != 'none') { m <- ctree(as.formula(paste('as.factor(',colnames(x)[par1],') ~ .',sep='')),data = x) if (par4=='yes') { a<-table.start() a<-table.row.start(a) a<-table.element(a,'10-Fold Cross Validation',3+2*par3,TRUE) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'',1,TRUE) a<-table.element(a,'Prediction (training)',par3+1,TRUE) a<-table.element(a,'Prediction (testing)',par3+1,TRUE) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'Actual',1,TRUE) for (jjj in 1:par3) a<-table.element(a,paste('C',jjj,sep=''),1,TRUE) a<-table.element(a,'CV',1,TRUE) for (jjj in 1:par3) a<-table.element(a,paste('C',jjj,sep=''),1,TRUE) a<-table.element(a,'CV',1,TRUE) a<-table.row.end(a) for (i in 1:10) { ind <- sample(2, nrow(x), replace=T, prob=c(0.9,0.1)) m.ct <- ctree(as.formula(paste('as.factor(',colnames(x)[par1],') ~ .',sep='')),data =x[ind==1,]) if (i==1) { m.ct.i.pred <- predict(m.ct, newdata=x[ind==1,]) m.ct.i.actu <- x[ind==1,par1] m.ct.x.pred <- predict(m.ct, newdata=x[ind==2,]) m.ct.x.actu <- x[ind==2,par1] } else { m.ct.i.pred <- c(m.ct.i.pred,predict(m.ct, newdata=x[ind==1,])) m.ct.i.actu <- c(m.ct.i.actu,x[ind==1,par1]) m.ct.x.pred <- c(m.ct.x.pred,predict(m.ct, newdata=x[ind==2,])) m.ct.x.actu <- c(m.ct.x.actu,x[ind==2,par1]) } } print(m.ct.i.tab <- table(m.ct.i.actu,m.ct.i.pred)) numer <- 0 for (i in 1:par3) { print(m.ct.i.tab[i,i] / sum(m.ct.i.tab[i,])) numer <- numer + m.ct.i.tab[i,i] } print(m.ct.i.cp <- numer / sum(m.ct.i.tab)) print(m.ct.x.tab <- table(m.ct.x.actu,m.ct.x.pred)) numer <- 0 for (i in 1:par3) { print(m.ct.x.tab[i,i] / sum(m.ct.x.tab[i,])) numer <- numer + m.ct.x.tab[i,i] } print(m.ct.x.cp <- numer / sum(m.ct.x.tab)) for (i in 1:par3) { a<-table.row.start(a) a<-table.element(a,paste('C',i,sep=''),1,TRUE) for (jjj in 1:par3) a<-table.element(a,m.ct.i.tab[i,jjj]) a<-table.element(a,round(m.ct.i.tab[i,i]/sum(m.ct.i.tab[i,]),4)) for (jjj in 1:par3) a<-table.element(a,m.ct.x.tab[i,jjj]) a<-table.element(a,round(m.ct.x.tab[i,i]/sum(m.ct.x.tab[i,]),4)) a<-table.row.end(a) } a<-table.row.start(a) a<-table.element(a,'Overall',1,TRUE) for (jjj in 1:par3) a<-table.element(a,'-') a<-table.element(a,round(m.ct.i.cp,4)) for (jjj in 1:par3) a<-table.element(a,'-') a<-table.element(a,round(m.ct.x.cp,4)) a<-table.row.end(a) a<-table.end(a) table.save(a,file='mytable3.tab') } } m bitmap(file='test1.png') plot(m) dev.off() bitmap(file='test1a.png') plot(x[,par1] ~ as.factor(where(m)),main='Response by Terminal Node',xlab='Terminal Node',ylab='Response') dev.off() if (par2 == 'none') { forec <- predict(m) result <- as.data.frame(cbind(x[,par1],forec,x[,par1]-forec)) colnames(result) <- c('Actuals','Forecasts','Residuals') print(result) } if (par2 != 'none') { print(cbind(as.factor(x[,par1]),predict(m))) myt <- table(as.factor(x[,par1]),predict(m)) print(myt) } bitmap(file='test2.png') if(par2=='none') { op <- par(mfrow=c(2,2)) plot(density(result$Actuals),main='Kernel Density Plot of Actuals') plot(density(result$Residuals),main='Kernel Density Plot of Residuals') plot(result$Forecasts,result$Actuals,main='Actuals versus Predictions',xlab='Predictions',ylab='Actuals') plot(density(result$Forecasts),main='Kernel Density Plot of Predictions') par(op) } if(par2!='none') { plot(myt,main='Confusion Matrix',xlab='Actual',ylab='Predicted') } dev.off() if (par2 == 'none') { detcoef <- cor(result$Forecasts,result$Actuals) a<-table.start() a<-table.row.start(a) a<-table.element(a,'Goodness of Fit',2,TRUE) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'Correlation',1,TRUE) a<-table.element(a,round(detcoef,4)) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'R-squared',1,TRUE) a<-table.element(a,round(detcoef*detcoef,4)) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'RMSE',1,TRUE) a<-table.element(a,round(sqrt(mean((result$Residuals)^2)),4)) a<-table.row.end(a) a<-table.end(a) table.save(a,file='mytable1.tab') a<-table.start() a<-table.row.start(a) a<-table.element(a,'Actuals, Predictions, and Residuals',4,TRUE) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'#',header=TRUE) a<-table.element(a,'Actuals',header=TRUE) a<-table.element(a,'Forecasts',header=TRUE) a<-table.element(a,'Residuals',header=TRUE) a<-table.row.end(a) for (i in 1:length(result$Actuals)) { a<-table.row.start(a) a<-table.element(a,i,header=TRUE) a<-table.element(a,result$Actuals[i]) a<-table.element(a,result$Forecasts[i]) a<-table.element(a,result$Residuals[i]) a<-table.row.end(a) } a<-table.end(a) table.save(a,file='mytable.tab') } if (par2 != 'none') { a<-table.start() a<-table.row.start(a) a<-table.element(a,'Confusion Matrix (predicted in columns / actuals in rows)',par3+1,TRUE) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'',1,TRUE) for (i in 1:par3) { a<-table.element(a,paste('C',i,sep=''),1,TRUE) } a<-table.row.end(a) for (i in 1:par3) { a<-table.row.start(a) a<-table.element(a,paste('C',i,sep=''),1,TRUE) for (j in 1:par3) { a<-table.element(a,myt[i,j]) } a<-table.row.end(a) } a<-table.end(a) table.save(a,file='mytable2.tab') }

Copyright

This work is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 License.

Software written by Ed van Stee & Patrick Wessa

Disclaimer

Information provided on this web site is provided "AS IS" without warranty of any kind, either express or implied, including, without limitation, warranties of merchantability, fitness for a particular purpose, and noninfringement. We use reasonable efforts to include accurate and timely information and periodically update the information, and software without notice. However, we make no warranties or representations as to the accuracy or completeness of such information (or software), and we assume no liability or responsibility for errors or omissions in the content of this web site, or any software bugs in online applications. Your use of this web site is AT YOUR OWN RISK. Under no circumstances and under no legal theory shall we be liable to you or any other person for any direct, indirect, special, incidental, exemplary, or consequential damages arising from your access to, or use of, this web site.

Privacy Policy

We may request personal information to be submitted to our servers in order to be able to:

• personalize online software applications according to your needs
• enforce strict security rules with respect to the data that you upload (e.g. statistical data)
• manage user sessions of online applications
• alert you about important changes or upgrades in resources or applications

We NEVER allow other companies to directly offer registered users information about their products and services. Banner references and hyperlinks of third parties NEVER contain any personal data of the visitor.

We do NOT sell, nor transmit by any means, personal information, nor statistical data series uploaded by you to third parties.

We carefully protect your data from loss, misuse, alteration, and destruction. However, at any time, and under any circumstance you are solely responsible for managing your passwords, and keeping them secret.

We store a unique ANONYMOUS USER ID in the form of a small 'Cookie' on your computer. This allows us to track your progress when using this website which is necessary to create state-dependent features. The cookie is used for NO OTHER PURPOSE. At any time you may opt to disallow cookies from this website - this will not affect other features of this website.

We examine cookies that are used by third-parties (banner and online ads) very closely: abuse from third-parties automatically results in termination of the advertising contract without refund. We have very good reason to believe that the cookies that are produced by third parties (banner ads) do NOT cause any privacy or security risk.

FreeStatistics.org is safe. There is no need to download any software to use the applications and services contained in this website. Hence, your system's security is not compromised by their use, and your personal data - other than data you submit in the account application form, and the user-agent information that is transmitted by your browser - is never transmitted to our servers.

As a general rule, we do not log on-line behavior of individuals (other than normal logging of webserver 'hits'). However, in cases of abuse, hacking, unauthorized access, Denial of Service attacks, illegal copying, hotlinking, non-compliance with international webstandards (such as robots.txt), or any other harmful behavior, our system engineers are empowered to log, track, identify, publish, and ban misbehaving individuals - even if this leads to ban entire blocks of IP addresses, or disclosing user's identity.

FreeStatistics.org is powered by