Merge pull request #37 from larahabashy/main

updating plot results

src/eda_cred.r

@@ -1,9 +1,9 @@
 # author: Lara Habashy
 # date: 2020-11-26
 
-"Creates eda plots for the pre-processed training data from the Wisconsin breast cancer data (from http://mlr.cs.umass.edu/ml/machine-learning-databases/breast-cancer-wisconsin/wdbc.data).
-Saves the plots as a pdf and png file.
-Usage: src/eda_cred.r --train=<train> --out_dir=<out_dir>
+"Creates eda plots for the pre-processed training data from the default of credit card clients data (from http://archive.ics.uci.edu/ml/datasets/default+of+credit+card+clients).
+Saves the plots as a png file.
+Usage: Rscript src/eda_cred.r --train=<train> --out_dir=<out_dir>
   
 Options:
 --train=<train>     Path (including filename) to training data (which needs to be saved as a feather file)
@@ -12,9 +12,12 @@ Options:
 
 library(feather)
 library(tidyverse)
-library(caret)
+#library(caret)
 library(docopt)
 library(ggthemes)
+#library (PerformanceAnalytics)
+#library(graphics)
+
 theme_set(theme_minimal())
 
 opt <- docopt(doc)
@@ -32,11 +35,41 @@ main <- function(train, out_dir) {
     xlab("Credit Limit") +
     ylab("Density") +
     ggtitle("Density of Credit Limit")
-    
+
   ggsave(paste0(out_dir, "/density_plot.png"), 
          density_plot,
          width = 8, 
          height = 10)
+
+  training_scaled$education <- as.factor(toupper(names(training_scaled[,23:29])[max.col(training_scaled[,23:29])]))
+
+  #histogram
+  education_histogram <- ggplot(training_scaled, aes(pay_1, fill = default)) + 
+    geom_histogram(binwidth = 1) + 
+    ggtitle("Histogram of Education Levels") +
+    facet_grid(.~education) + 
+    theme_fivethirtyeight() 
+
+  ggsave(paste0(out_dir, "/education_histogram.png"), 
+         education_histogram,
+         width = 8, 
+         height = 10)
+
+  numeric_df <- training_scaled
+  numeric_df$default <- as.numeric(numeric_df$default)
+  numeric_df$age <- as.numeric(numeric_df$age)
+  numeric_df$sex <- NULL
+  numeric_df$education <- NULL
+  numeric_df$marriage <- NULL
+  ggsave(corr_plot, "correlation_plot.png")
+  #correlation plot
+  corr_plot <- PerformanceAnalytics::chart.Correlation(numeric_df %>% select_if(is.numeric), histogram=TRUE, method = "pearson", col="blue", pch=1, main="all")
+
+  ggsave(paste0(out_dir, "/correlation_plot.png"), 
+         corr_plot,
+         width = 8, 
+         height = 10)
+
 }
 
 main(opt[["--train"]], opt[["--out_dir"]])