Compare LM outputs and visualize startprice variation acrocss biddable/sold-combinations

comparison.R

library(ggplot2)
library(dplyr)
library(readr)

ebay_train <- read.csv("./data/eBayiPadTrain.csv")

# Remove observations which cause trouble for later predictions
ebay_train <- ebay_train %>% filter(carrier != 'Other' | productline != 'iPad 5')


bid_sold_all <- ebay_train %>% select(biddable, startprice, condition, cellular
                                      carrier, color, storage, productline)

# Do not select biddable because it was used for subetting already
b0_s1 <- ebay_train %>% filter(biddable == 0, sold == 1) %>%  
  select(startprice, condition, cellular, color, carrier
         storage, productline)

# Do not select biddable because it was used for subetting already
b0s1_b0s1_b1s0 <- ebay_train %>% filter(biddable != 1, sold != 1) %>%
  select(startprice, condition, cellular, color, carrier
         storage, productline)
  


lm_all <- lm(startprice ~ ., data = bid_sold_all)
summary(lm_all)

lm_b0s1 <- lm(startprice ~ ., data = b0_s1)
summary(lm_b0s1)

lm_b0s1_b0s1_b1s0 <- lm(startprice ~ ., data = b0s1_b0s1_b1s0)
summary(lm_b0s1_b0s1_b1s0)


ebay_train$price_diff_all <- ebay_train$startprice - predict(lm_all, 
                                                             newdata = ebay_train)

ebay_train$price_diff_b0s1 <- ebay_train$startprice - predict(lm_b0s1, 
                                                              newdata = ebay_train)

ebay_train$price_diff_b0s1_b0s1_b1s0 <- ebay_train$startprice - predict(lm_b0s1_b0s1_b1s0, 
                                                                        newdata = ebay_train)


# Modify entries for plotting later
ebay_train$sold <- ifelse(ebay_train$sold == 0, 'not sold', 'sold')
ebay_train$biddable <- ifelse(ebay_train$biddable == 0, 'non-biddable', 'biddable')


qplot(productline, price_diff_all, data = ebay_train, geom = "boxplot", 
      facets = biddable ~ sold)

qplot(productline, price_diff_b0s1, data = ebay_train, geom = "boxplot", 
      facets = biddable ~ sold)

qplot(productline, price_diff_b0s1_b0s1_b1s0, data = ebay_train, geom = "boxplot", 
      facets = biddable ~ sold)


qplot(storage, price_diff_all, data = ebay_train, geom = "boxplot", 
      facets = biddable ~ sold)

qplot(storage, price_diff_b0s1, data = ebay_train, geom = "boxplot", 
      facets = biddable ~ sold)

qplot(storage, price_diff_b0s1_b0s1_b1s0, data = ebay_train, geom = "boxplot", 
      facets = biddable ~ sold)

qplot(condition, price_diff_all, data = ebay_train, geom = "boxplot", 
      facets = biddable ~ sold)

qplot(condition, price_diff_b0s1, data = ebay_train, geom = "boxplot", 
      facets = biddable ~ sold)

qplot(condition, price_diff_b0s1_b0s1_b1s0, data = ebay_train, geom = "boxplot", 
      facets = biddable ~ sold)