SUSA-org · JasonAsh00 · Apr 15, 2018 · Apr 19, 2018 · May 2, 2018 · May 2, 2018
diff --git a/Kaggle/DATA/house-prices/test.csv → Kaggle/DATA/house-prices/final_test.csv b/Kaggle/DATA/house-prices/test.csv → Kaggle/DATA/house-prices/final_test.csv
diff --git a/Kaggle/DATA/house-prices/full.csv b/Kaggle/DATA/house-prices/full.csv
diff --git a/Kaggle/DATA/house-prices/test_cleaned.csv b/Kaggle/DATA/house-prices/test_cleaned.csv
diff --git a/Kaggle/DATA/house-prices/train_cleaned.csv b/Kaggle/DATA/house-prices/train_cleaned.csv
diff --git a/Kaggle/Kshitiz Gutpa Spring 2018 Resume.docx b/Kaggle/Kshitiz Gutpa Spring 2018 Resume.docx
diff --git a/Kaggle/feature_eng.r b/Kaggle/feature_eng.r
@@ -0,0 +1,23 @@
+library(ggplot2); library(dplyr): library(readr)
+
+# read in the data
+
+clean <- read_r('train_cleaned.csv')
+
+
+num_featrs <- c('LotFrontage', 'LotArea' ,'YearBuilt', 'YearRemodAdd',
+                'MasVnrArea','BsmtFinSF1','BsmtUnfSF', 'TotalBsmtSF','1stFlrSF',
+                '2ndFlrSF','GrLivArea',
+                'TotRmsAbvGrd','GarageArea', 'WoodDeckSF', 'OpenPorchSF',
+                'EnclosedPorch','PoolArea','MiscValue')
+
+y <- select(poly_data, 'SalePrice')
+scatter_plots <- function(df, feat) {
+  df  %>% 
+    ggplot(aes(feat, df$SalePrice)) + facet_wrap(~ feat, scales = "free") + 
+    theme(legend.position = "none") + labs(x = "", y = "")
+}
+
+box_plots <- function(df) {
+
+}