A teacher might score like this …

Load Data Set

# Load Roster Data
load("./RData/OriginalRoster.RData")
print(roster)

##              Student Math Science English
## 1         John Davis  502      95      25
## 2    Angela Williams  600      99      22
## 3   Bullwinkle Moose  412      80      18
## 4        David Jones  358      82      15
## 5  Janice Markhammer  495      75      20
## 6     Cheryl Cushing  512      85      28
## 7     Reuven Ytzrhak  410      80      15
## 8          Greg Knox  625      95      30
## 9       Joel England  573      89      27
## 10      Mary Rayburn  522      86      18

Modeling

z <- scale(roster[, 2:4])  # standardize different scales
score <- apply(z, 1, mean)  # take means
roster <- cbind(roster, score)  # combine with original roster
print(roster)

##              Student Math Science English   score
## 1         John Davis  502      95      25  0.5592
## 2    Angela Williams  600      99      22  0.9238
## 3   Bullwinkle Moose  412      80      18 -0.8565
## 4        David Jones  358      82      15 -1.1620
## 5  Janice Markhammer  495      75      20 -0.6290
## 6     Cheryl Cushing  512      85      28  0.3532
## 7     Reuven Ytzrhak  410      80      15 -1.0476
## 8          Greg Knox  625      95      30  1.3379
## 9       Joel England  573      89      27  0.6978
## 10      Mary Rayburn  522      86      18 -0.1768

Evaluation

y <- quantile(score, c(0.8, 0.6, 0.4, 0.2))  # set criteria
print(y)

##     80%     60%     40%     20% 
##  0.7430  0.4356 -0.3577 -0.8948

roster$grade[score >= y[1]] <- "A"  # grading
roster$grade[score < y[1] & score >= y[2]] <- "B"
roster$grade[score < y[2] & score >= y[3]] <- "C"
roster$grade[score < y[3] & score >= y[4]] <- "D"
roster$grade[score < y[4]] <- "F"

Performance

name <- strsplit(as.character(roster$Student), " ")
lastname <- sapply(name, "[", 2)
firstname <- sapply(name, "[", 1)
roster <- cbind(`First Name` = firstname, `Last Name` = lastname, roster[, -1])
roster <- roster[order(score, decreasing = TRUE), ]
# Score Report
(click on any column header to sort table automatically! )

A researcher might score like this …

Load Data Set

# Load German Credit data
require(caret)
data(GermanCredit)
dim(GermanCredit)[1]  # No. of obs

## [1] 1000

dim(GermanCredit)[2]  # No. of variables (original)

## [1] 62

Modeling (1)

# Build a container for German Credit Data Set
gcDS = new.env()

# check container data after clean-up and partitions

# training data
length(attributes(gcDS$train)$names)  # No. of variables

## [1] 22

length(attributes(gcDS$train)$row.names)  # No. of obs

## [1] 700

# testing data
length(attributes(gcDS$test)$names)  # No. of variables

## [1] 22

length(attributes(gcDS$test)$row.names)  # No. of obs

## [1] 300

Modeling (2)

# Build a container for Generalized Linear Model (GLM)
gcGLM = new.env(parent = gcDS)

# Build a container for Random Forest Model (RFOREST)
require(randomForest)
gcRFOREST = new.env(parent = gcDS)

# Build a container for Support Vector Machines Model (SVM)
require(e1071)
gcSVM = new.env(parent = gcDS)

# Build a container for 1-hidden-layer Neural Network Model (NNET)
require(nnet)
gcNNET = new.env(parent = gcDS)

# Build a container for Stochastic Boosting Model (BOOST)
require(ada)
gcBOOST = new.env(parent = gcDS)

# Build a container for Naive Bayes Classifier Model (NBayes)
require(e1071)
gcNBayes = new.env(parent = gcDS)

# Build a container for k-Nearest Neighbour Classification Model (KNN)
require(class)
gcKNN = new.env(parent = gcDS)

# Build a container for Linear Discriminant Analysis Model (LDA)
require(MASS)
gcLDA = new.env(parent = gcDS)

Evaluation (1)

# Area Under the ROC Curve (AUC)

# ROC curves

Evaluation (2)

# Given Cost Matrix (Actual vs. Predicted)

# Cost Curves

Performance

# Risk Chart

Modeling in R, Scoring in SAS chosen for ‘that’ Big Data

Verifying their scores for comparison.
(click on any column header to sort table automatically! )

To me, a teacher and a researcher are one.

my sessionInfo …

print(sessionInfo(), locale = FALSE)

## R version 2.15.2 (2012-10-26)
## Platform: i386-w64-mingw32/i386 (32-bit)
## 
## attached base packages:
## [1] grid      stats     graphics  grDevices utils     datasets  methods  
## [8] base     
## 
## other attached packages:
##  [1] boot_1.3-7         vcd_1.2-13         colorspace_1.2-1  
##  [4] RColorBrewer_1.0-5 scales_0.2.3       ggplot2_0.9.3     
##  [7] ROCR_1.0-4         gplots_2.11.0      MASS_7.3-23       
## [10] KernSmooth_2.23-8  caTools_1.14       gdata_2.12.0.2    
## [13] gtools_2.7.0       ada_2.0-3          rpart_4.1-0       
## [16] nnet_7.3-5         e1071_1.6-1        class_7.3-5       
## [19] randomForest_4.6-7 caret_5.15-61      reshape2_1.2.2    
## [22] plyr_1.8           lattice_0.20-13    foreach_1.4.0     
## [25] cluster_1.14.3     knitr_1.0.5       
## 
## loaded via a namespace (and not attached):
##  [1] bitops_1.0-5    codetools_0.2-8 dichromat_2.0-0 digest_0.6.3   
##  [5] evaluate_0.4.3  formatR_0.7     gtable_0.1.2    iterators_1.0.6
##  [9] labeling_0.1    munsell_0.4     proto_0.3-10    stringr_0.6.2  
## [13] tools_2.15.2