Loading database
library(tidyverse)
# loading database without index and desc column
df <- read.csv("BigBasket Products.csv") %>% select(-index,-description)
# factor distinct columns
df <- df %>% mutate(category=as_factor(category),
sub_category=as_factor(sub_category),
brand=as_factor(brand),
type=as_factor(type))
# adding discount column
df <- df %>% mutate(discount=round((market_price-sale_price)/market_price, digits = 3))
Summary the data
summary(df)
## product category
## Length:27555 Beauty & Hygiene :7867
## Class :character Gourmet & World Food :4690
## Mode :character Kitchen, Garden & Pets :3580
## Snacks & Branded Foods :2814
## Foodgrains, Oil & Masala:2676
## Cleaning & Household :2675
## (Other) :3253
## sub_category brand sale_price
## Skin Care : 2294 Fresho : 638 Min. : 2.45
## Health & Medicine : 1133 bb Royal : 539 1st Qu.: 95.00
## Hair Care : 1028 BB Home : 428 Median : 190.00
## Storage & Accessories: 1015 DP : 250 Mean : 322.51
## Fragrances & Deos : 1000 Fresho Signature: 171 3rd Qu.: 359.00
## Bath & Hand Wash : 996 bb Combo : 168 Max. :12500.00
## (Other) :20089 (Other) :25361
## market_price type rating
## Min. : 3.0 Face Care : 1508 Min. :1.000
## 1st Qu.: 100.0 Ayurveda : 538 1st Qu.:3.700
## Median : 220.0 Men's Deodorants : 500 Median :4.100
## Mean : 382.1 Shampoo & Conditioner: 461 Mean :3.943
## 3rd Qu.: 425.0 Containers Sets : 415 3rd Qu.:4.300
## Max. :12500.0 Glassware : 415 Max. :5.000
## (Other) :23718 NA's :8626
## discount
## Min. :0.0000
## 1st Qu.:0.0000
## Median :0.0500
## Mean :0.1182
## 3rd Qu.:0.2000
## Max. :0.8370
##
% of NA’s in rating category
df %>% summarise(perc_of_na_rating = sum(is.na(rating))/n()*100)
## perc_of_na_rating
## 1 31.30466
About 31% of rating data don’t have value.
Data structure
str(df)
## 'data.frame': 27555 obs. of 9 variables:
## $ product : chr "Garlic Oil - Vegetarian Capsule 500 mg" "Water Bottle - Orange" "Brass Angle Deep - Plain, No.2" "Cereal Flip Lid Container/Storage Jar - Assorted Colour" ...
## $ category : Factor w/ 11 levels "Beauty & Hygiene",..: 1 2 3 3 1 3 1 1 1 3 ...
## $ sub_category: Factor w/ 90 levels "Hair Care","Storage & Accessories",..: 1 2 3 4 5 6 7 5 1 8 ...
## $ brand : Factor w/ 2314 levels "Sri Sri Ayurveda ",..: 1 2 3 4 5 6 7 8 9 10 ...
## $ sale_price : num 220 180 119 149 162 ...
## $ market_price: num 220 180 250 176 162 ...
## $ type : Factor w/ 426 levels "Hair Oil & Serum",..: 1 2 3 4 5 6 7 8 9 10 ...
## $ rating : num 4.1 2.3 3.4 3.7 4.4 3.3 3.6 4 3.5 4.3 ...
## $ discount : num 0 0 0.524 0.153 0 0.151 0 0 0 0 ...
The most popular rating is about 4.
df %>% ggplot(aes(x=rating, fill=..x..))+
geom_histogram(show.legend = F)+
scale_fill_gradient(low="red", high="green")
What is the share of each category
df %>%
group_by(category) %>%
summarise(count=n()) %>%
mutate(perc=count/sum(count)) %>%
arrange(count)%>%
ggplot(aes(y=fct_reorder(category, count), x=perc*100, fill=category))+
geom_col(show.legend = F)+
labs(x="Percent", y="", title = "Percentage of a category")+
theme_minimal()
About 30% items are from Beauty category.
Which brands have the best rating?
df2 <- df %>% group_by(brand) %>%
filter(!is.na(rating)) %>%
summarise(mean_rating = mean(rating), n_element = n()) %>%
arrange(desc(mean_rating))
# top rating brands which have more than 4 products
df2 %>% filter(n_element>4) %>% head(10) %>% select(-n_element)
## # A tibble: 10 x 2
## brand mean_rating
## <fct> <dbl>
## 1 Sangam Sweets 4.83
## 2 Kaadoo 4.81
## 3 Bobs Red Mill 4.79
## 4 Nimyle 4.78
## 5 Continent Spice Khadi 4.76
## 6 Pine Sol 4.76
## 7 The Whole Truth 4.75
## 8 SOVI 4.66
## 9 Pyrex 4.63
## 10 Tabasco 4.62
# top rating brands which have more than 9 products
df2 %>% filter(n_element>9) %>% head(10) %>% select(-n_element)
## # A tibble: 10 x 2
## brand mean_rating
## <fct> <dbl>
## 1 "Hi-Luxe" 4.57
## 2 "VAHDAM" 4.49
## 3 "Morpheme Remedies" 4.46
## 4 "Papergrid" 4.43
## 5 "Whiskas" 4.42
## 6 "Parachute " 4.41
## 7 "Tide" 4.4
## 8 "BB Home Herbal" 4.39
## 9 "Laopala Diva" 4.38
## 10 "Selpak" 4.38
Discounts for each category?
df %>% ggplot(aes(y=category, x=discount, color=category)) +
geom_boxplot(show.legend=F)
The vast majority of discounts are from 0 to 20%.
We can see that there are suspiciously many values around 0.2 in the
Fruits & Vegetables category.
df %>% filter(category=="Fruits & Vegetables") %>%
summarise(discout_equal_0.2 =sum(discount==0.2)/n())
## discout_equal_0.2
## 1 0.8743268
87% of products from the Fruits & Vegetables category have 20%
discount.
Does discounting affect product rating?
df %>% na.omit(rating) %>%
ggplot(aes(discount, rating))+
geom_point()+
geom_smooth()+
facet_wrap(vars(category))
There is huge spread and it seems that discounting doesn’t affect
rating. For each category rating is about 4.