search

naomihng / coupon-purchase-prediction

Predict coupon purchase based on past purchase and browsing behavior (data from Kaggle)
1 stars 1 forks source link

discount rate vs age group / gender #7

Open naomihng opened 7 years ago

cherylthm commented 7 years ago

Find discount rate vs age group / gender

Load coupon_detail, coupon_list, user_list

library(ggplot2)

coupon_detail_train = read.csv("coupon_detail_train.csv", stringsAsFactors = F)
coupon_list_train = read.csv("coupon_list_train_en.csv", stringsAsFactors = F)
user_list = read.csv("user_list_en.csv", stringsAsFactors = F)

str(coupon_detail_train)
str(coupon_list_train)
str(user_list)

Isolate USER_ID_hash and COUPON_ID_hash from coupon_detail_train

user.coupon.id = data.frame(USER_ID_hash=coupon_detail_train$USER_ID_hash, COUPON_ID_hash=coupon_detail_train$COUPON_ID_hash)
#Isolate PRICE_RATE and COUPON_ID_hash from coupon_list_train
coupon_disc_rate = data.frame(COUPON_ID_hash=coupon_list_train$COUPON_ID_hash,PRICE_RATE=coupon_list_train$PRICE_RATE)
#Isolate SEX_ID, AGE, and USER_ID_hash from user_list
user_demo = data.frame(USER_ID_hash=user_list$USER_ID_hash, SEX_ID= user_list$SEX_ID, AGE=user_list$AGE)

#Merge user.coupon.id with coupon_disc_rate and user_demo
disc.rate.corr = merge(user.coupon.id, coupon_disc_rate, by="COUPON_ID_hash", all.x=T)
disc.rate.corr = merge(disc.rate.corr, user_demo, by="USER_ID_hash", all.x=T)

Discount rate v Age

ggplot(disc.rate.corr,aes(x=AGE,y=PRICE_RATE)) + geom_point(fill='blue') + ylim(0,101) + stat_smooth(method="lm") 
#No correlation

Discount rate v Gender

ggplot(disc.rate.corr) + geom_boxplot(aes(x=SEX_ID,y=PRICE_RATE))

Scatterplot of discount rate v age, + gender

ggplot(disc.rate.corr,aes(x=AGE,y=PRICE_RATE, color=SEX_ID)) + geom_point() + ylim(0,101) + stat_smooth(method="lm") 
#No correlation