Update plot titles; add more printout messages; load ggplot2 in linear reg

examples/MRS_and_Machine_Learning/Machine Learning/Gradient Boosting Machine.R

@@ -6,9 +6,9 @@
 # ----------------------------------------------------------------------------
 # load packages
 # ----------------------------------------------------------------------------
-(if (!require("MASS")) install.packages("MASS"))
+(if (!require("MASS", quietly = TRUE)) install.packages("MASS"))
 library("MASS") # to use the Boston dataset
-(if (!require("gbm")) install.packages("gbm"))
+(if (!require("gbm", quietly = TRUE)) install.packages("gbm"))
 library("gbm") # Gradient Boosting Machine package
 
 # ----------------------------------------------------------------------------

examples/MRS_and_Machine_Learning/Machine Learning/LASSO Model.R

@@ -9,9 +9,9 @@
 # ----------------------------------------------------------------------------
 # load packages
 # ----------------------------------------------------------------------------
-(if (!require("glmnet")) install.packages("glmnet"))
+(if (!require("glmnet", quietly = TRUE)) install.packages("glmnet"))
 library("glmnet") # use this package to fit a glmnet model
-(if (!require("MASS")) install.packages("MASS"))
+(if (!require("MASS", quietly = TRUE)) install.packages("MASS"))
 library("MASS") # to use the Boston dataset
 
 # ----------------------------------------------------------------------------

examples/MRS_and_Machine_Learning/Machine Learning/Linear Regression and Azure Web Service.R

@@ -15,10 +15,13 @@ auth_token <- ""
 # ----------------------------------------------------------------------------
 # load packages
 # ----------------------------------------------------------------------------
-(if (!require("AzureML")) install.packages("AzureML"))
+# install packages if they are not already installed
+(if (!require("AzureML", quietly = TRUE)) install.packages("AzureML"))
 library("AzureML") # load the package for deploying Azure ML web service
-(if (!require("MASS")) install.packages("MASS"))
+(if (!require("MASS", quietly = TRUE)) install.packages("MASS"))
 library("MASS") # to use the Boston dataset
+if (!require("ggplot2", quietly = TRUE)) install.packages("ggplot2")
+library("ggplot2") # used for plotting
 
 # ----------------------------------------------------------------------------
 # fit a model and check model performance
@@ -26,7 +29,7 @@ library("MASS") # to use the Boston dataset
 # check the data
 head(Boston)
 ggplot(Boston, aes(x=medv)) + 
-  geom_histogram(binwidth=5) +
+  geom_histogram(binwidth=2) +
   ggtitle("Histogram of Response Variable")
 
 # fit a model using medv as response and others as predictors 

examples/MRS_and_Machine_Learning/R_MRO_MRS_Comparison/Part_1a_Functions_glm_rxGlm.R

@@ -25,10 +25,7 @@ if (!RRE)
 }
 
 # install a package if it's not already installed
-if (!require("ggplot2", quietly = TRUE))
-  install.packages("ggplot2")
-
-# load libraries
+if (!require("ggplot2", quietly = TRUE)) install.packages("ggplot2")
 library("ggplot2") # used for plotting
 
 # ----------------------------------------------------------------------------

examples/MRS_and_Machine_Learning/R_MRO_MRS_Comparison/Part_1b_Functions_kmeans_rxKmeans.R

@@ -25,8 +25,7 @@ if (!RRE)
 }
 
 # install a package if it's not already installed
-if (!require("ggplot2", quietly = TRUE))
-  install.packages("ggplot2")
+if (!require("ggplot2", quietly = TRUE)) install.packages("ggplot2")
 
 # load packages
 library("MASS") # to use the mvrnorm function
@@ -63,7 +62,7 @@ ggplot(group_all, aes(x = V1, y = V2)) +
   xlim(-5, 5) + ylim(-5, 5) +
   geom_hline(yintercept = 0) +
   geom_vline(xintercept = 0) +
-  ggtitle("Simulated data in two overlapping groups")
+  ggtitle("Simulated Data in Two Overlapping Groups")
 
 # assign data 
 mydata <- group_all[, 1:2]
@@ -88,7 +87,7 @@ ggplot(mydata_clusters, aes(x = V1, y = V2)) +
   xlim(-5, 5) + ylim(-5, 5) +
   geom_hline(yintercept = 0) +
   geom_vline(xintercept = 0) +
-  ggtitle("Clusters found by kmeans()")
+  ggtitle("Clusters Found by kmeans()")
 
 # ----------------------------------------------------------------------------
 # cluster analysis with rxKmeans(), it works on MRS only
@@ -128,7 +127,7 @@ if (RRE){
     xlim(-5, 5) + ylim(-5, 5) +
     geom_hline(yintercept = 0) +
     geom_vline(xintercept = 0) +
-    ggtitle("Clusters found by rxKmeans()")
+    ggtitle("Clusters Found by rxKmeans()")
   
 } else{
   print("rxKmeans was not run becauase the RevoScaleR package is not available")

examples/MRS_and_Machine_Learning/R_MRO_MRS_Comparison/Part_2_Capacity.R

@@ -25,8 +25,7 @@ if (RRE)
 }
 
 # install a package if it's not already installed
-if (!require("ggplot2", quietly = TRUE))
-  install.packages("ggplot2")
+if (!require("ggplot2", quietly = TRUE)) install.packages("ggplot2")
 
 # load packages
 library("MASS") # to use the mvrnorm function
@@ -63,14 +62,14 @@ group_all <- rbind(group_a, group_b)
 nclusters <- 2
 
 # plot sample data
-plot_data <- group_all[sample(nrow(group_all), 1000),] 
+plot_data <- group_all[sample(2 * nsamples, min(1000, 2 * nsamples)),] 
 ggplot(plot_data, aes(x = V1, y = V2)) +
   geom_point(aes(colour = group)) +
   geom_point(data = data.frame(V1 = c(-1, 1), V2 = c(-1, 1)), size = 5) +
   xlim(-5, 5) + ylim(-5, 5) +
   geom_hline(yintercept = 0) +
   geom_vline(xintercept = 0) +
-  ggtitle("Simulated data in two overlapping groups")
+  ggtitle("Simulated Data in Two Overlapping Groups")
 
 # save data
 mydata = group_all[, 1:2]

examples/MRS_and_Machine_Learning/R_MRO_MRS_Comparison/Part_3a_Speed_for_Matrix_Calculations.R

@@ -51,3 +51,5 @@ k <- round(m / 2)
 A <- data.frame(A, fac = sample(LETTERS[1:g], m, replace = TRUE))
 train <- sample(1:m, k)
 system.time(L <- lda(fac ~ ., data = A, prior = rep(1, g) / g, subset = train))
+
+message("Save the time and run the code on R, MRO and MRS to compare speed.")

examples/MRS_and_Machine_Learning/R_MRO_MRS_Comparison/Part_3b_Speed_for_kmeans.R

@@ -38,7 +38,12 @@ group_all <- rbind(group_a, group_b)
 nclusters <- 2
 
 mydata = group_all[, 1:2]
+
+message("It might take a while for this to finish if nsamples is large.")
 # K-Means Cluster Analysis
 system_time_r <- system.time(fit <- kmeans(mydata, nclusters,
                                            iter.max = 1000,
                                            algorithm = "Lloyd"))
+system_time_r
+
+message("Save the time and run the code on R, MRO and MRS to compare speed.")

examples/MRS_and_Machine_Learning/R_MRO_MRS_Comparison/Part_3c_Speed_for_kmeans_rxKmeans.R

@@ -25,8 +25,7 @@ if (!RRE)
 }
 
 # install a package if it's not already installed
-if (!require("ggplot2", quietly = TRUE))
-  install.packages("ggplot2")
+if (!require("ggplot2", quietly = TRUE)) install.packages("ggplot2")
 
 # load libraries
 library("MASS") # to use the mvrnorm function
@@ -54,6 +53,9 @@ simulCluster <- function(nsamples, mean, dimension, group)
   z
 }
 
+message("It might take a while for this to finish if any of the elements in ", 
+  "nsamples_list is large.")
+
 for (nsamples in nsamples_list)
 {
   # simulate data and append
@@ -72,7 +74,8 @@ for (nsamples in nsamples_list)
   # kmeans with MRS
   
   if (RRE){
-    system_time_rre <- system.time(clust <- rxKmeans( ~ V1 + V2, data = mydata,
+    system_time_rre <- system.time(clust <- rxKmeans( ~ V1 + V2, 
+                                                      data = mydata,
                                                       numClusters = nclusters,
                                                       algorithm = "lloyd"))
   }
@@ -99,10 +102,12 @@ if (RRE){
     geom_point(aes(y = time_rre, colour = "rxKmeans")) +
     geom_line(aes(y = time_rre, colour = "rxKmeans")) +
     scale_x_continuous(breaks = seq(2, 8, by = 1)) +
-    scale_colour_manual("Function", values = c(kmeans = "red", rxKmeans = "blue")) +
+    scale_colour_manual("Function", 
+                        values = c(kmeans = "red", rxKmeans = "blue")) +
     xlab("log10(number of samples)") +
     ylab("time in seconds") +
-    ggtitle("If data fits in memory, kmeans() and rxKmeans() are equally performant")
+    ggtitle(paste("If data fits in memory,", 
+                  "kmeans() and rxKmeans() are equally performant"))
 } else {
   ggplot(data = mydata, aes(x = nsamples_log)) +
     geom_point(aes(y = time_r, colour = "kmeans")) +
@@ -111,5 +116,6 @@ if (RRE){
     scale_colour_manual("Function", values = c(kmeans = "red")) +
     xlab("log10(number of samples)") +
     ylab("time in seconds") +
-    ggtitle("Time for kmeans. To add time for rxKmean, use the RRE engine")
+    ggtitle(paste("Time for kmeans \n", 
+                  "To add time for rxKmeans, use the R Server engine"))
 }