Q1) a. Creating a new variable age_group
nyt1 <- read.csv("C:\\Users\\welcome\\Desktop\\Stuff\\MEngg\\Sem-1\\DS\\nyt1.csv")
# Dividing Age into different categories
nyt1$age_group <- cut(nyt1$Age, c(0,18,25,35,45,55,65,Inf), right= FALSE, labels = c("<18","18-24","25-34","35-44","45-54","55-64","65+"))
# Summary
summary(nyt1)
## Age Gender Impressions Clicks
## Min. : 0.00 Min. :0.000 Min. : 0.000 Min. :0.00000
## 1st Qu.: 0.00 1st Qu.:0.000 1st Qu.: 3.000 1st Qu.:0.00000
## Median : 31.00 Median :0.000 Median : 5.000 Median :0.00000
## Mean : 29.48 Mean :0.367 Mean : 5.007 Mean :0.09259
## 3rd Qu.: 48.00 3rd Qu.:1.000 3rd Qu.: 6.000 3rd Qu.:0.00000
## Max. :108.00 Max. :1.000 Max. :20.000 Max. :4.00000
##
## Signed_In age_group
## Min. :0.0000 <18 :150934
## 1st Qu.:0.0000 18-24: 40694
## Median :1.0000 25-34: 58174
## Mean :0.7009 35-44: 70860
## 3rd Qu.:1.0000 45-54: 64288
## Max. :1.0000 55-64: 44738
## 65+ : 28753
Q1) b. (i) Plotting Impressions and CTR for 6 categories
library(ggplot2)
#CTR vs Impression Graph
g1 <- ggplot(nyt1, aes(x=Impressions,y=((Clicks/Impressions)),color= age_group)) +geom_point(size = 2) + labs(title="CTR vs Impressions")
g1 + labs(y="CTR")
## Warning: Removed 3066 rows containing missing values (geom_point).
#Impressions
library(ggplot2)
b <- ggplot(nyt1, aes(Impressions, fill=age_group))
b+ geom_bar() + labs(title="Impressions")
#CTR
ctr <- ifelse(nyt1$Impressions !=0, nyt1$Clicks/nyt1$Impressions,NA)
nyt1$CTR <- ctr
c <- ggplot(data = nyt1,aes(CTR, fill = age_group, color = age_group))
c + geom_histogram(position = "dodge", binwidth = 0.1) + labs(x = "CTR",title="CTR")
## Warning: Removed 3066 rows containing non-finite values (stat_bin).
Q1)b.
(ii) New Variable to segment by click behaviour
#Plot to visualise clicks by different age groups
ggplot(nyt1, aes(y=age_group, x=Clicks, color=age_group)) + geom_line(size=2) + theme_bw()
Less
than 18 year segment recorded highest number of clicks by <18
followed by 55+ , 35-44 and other segments. So, I categorised with
the new variable "Click_Behavior". (Zero, Low, Medium and High) as
shown.
nyt1$Click_Behavior <- ifelse(is.na(nyt1$CTR), 'NA', ifelse(nyt1$CTR==0, 'Zero',ifelse(nyt1$CTR>0 & nyt1$CTR<=0.2, 'Low',
ifelse(nyt1$CTR>0.2 & nyt1$CTR<=0.4, 'Medium',
'High'))))
summary(nyt1$Click_Behavior)
## Length Class Mode
## 458441 character character
Q1) b. (iii) Visual Comparisions
library(ggplot2)
segment <- ifelse(nyt1$Age < 18,c("<18"), c(">=18"))
nyt1$Segment <- segment
d<- ggplot(nyt1, aes(y=CTR, x=Impressions, color=factor(Gender))) + geom_point(data=subset(nyt1, Segment=='<18'),size=2) + labs(title="<18 Male vs Female", y="CTR", color="Gender")
d + geom_point(data=subset(nyt1, Segment=='>=18'),size=2) + labs(title="<18 Male vs Female and >=18 Male and Female Clicks", y="CTR", color="Gender") +scale_color_manual(labels = c("Female", "Male"), values = c("blue", "red")) + facet_grid(.~ Segment)
## Warning: Removed 1025 rows containing missing values (geom_point).
## Warning: Removed 2041 rows containing missing values (geom_point).
Q1) C. Potential Metrics and Weekly Trend
#As we estimated female under 18 recorded highest number of clicks, taking their mean would estimate their age on an average. (Taking age>0)
mean(nyt1[nyt1$Age>0 & nyt1$Age<18 & nyt1$Gender==0,"Age"])
## [1] 15.24621
var(nyt1[nyt1$Age>0 & nyt1$Age<18 & nyt1$Gender==0, "Age"])
## [1] 2.914118
15 years is the average female age with variance 3 recording highest number of clicks. The max value is 17.
mean(nyt1[nyt1$Age>=18 & nyt1$Gender==1,"Age"])
## [1] 42.34767
median(nyt1[nyt1$Age>=18 & nyt1$Gender==1,"Age"])
## [1] 41
*As Men recorded highest number of clicks in the segment >=18 years are 42 years old on an average. Median is 41 (distribution will be skewed to the right) *
#R Code for visualizing weekly trend
nyt1 <- read.csv("C:\\Users\\welcome\\Desktop\\Stuff\\MEngg\\Sem-1\\DS\\nyt1.csv")
nyt2 <- read.csv("C:\\Users\\welcome\\Desktop\\Stuff\\MEngg\\Sem-1\\DS\\nyt2.csv")
nyt3 <- read.csv("C:\\Users\\welcome\\Desktop\\Stuff\\MEngg\\Sem-1\\DS\\nyt3.csv")
nyt4 <- read.csv("C:\\Users\\welcome\\Desktop\\Stuff\\MEngg\\Sem-1\\DS\\nyt4.csv")
nyt5 <- read.csv("C:\\Users\\welcome\\Desktop\\Stuff\\MEngg\\Sem-1\\DS\\nyt5.csv")
nyt6 <- read.csv("C:\\Users\\welcome\\Desktop\\Stuff\\MEngg\\Sem-1\\DS\\nyt6.csv")
nyt7 <- read.csv("C:\\Users\\welcome\\Desktop\\Stuff\\MEngg\\Sem-1\\DS\\nyt7.csv")
nyt1$day <- "day1"
nyt2$day <- "day2"
nyt3$day <- "day3"
nyt4$day <- "day4"
nyt5$day <- "day5"
nyt6$day <- "day6"
nyt7$day <- "day7"
nyt <- rbind(nyt1, nyt2, nyt3, nyt4, nyt5, nyt6, nyt7)
levels(nyt$day) <- list('day1' = 'day1',
'day2' = 'day2',
'day3' = 'day3',
'day4' = 'day4',
'day5' = 'day5',
'day6' = 'day6',
'day7' = 'day7')
nyt$Gender <- ifelse(nyt$Gender == 0, 'Female','Male')
week <- ggplot(nyt, aes(x=day, y=Clicks))
week + geom_bar(stat = "identity") + facet_grid(.~ Gender)
- Less than 18 segment record the highest CTR among all the age groups. Female under 18 have the highest clicks than females who are more than 18 years old. Males who are above 18 years record more clicks than who are less than 18 years and Males record more number of clicks than female in the above 18 segment having an average age of 42 years.
- However, the CTR rate for most of male and female is less than 0.2.
- In Weekly trend, it is clear that overall female clicks are higher than male. It is also evident that Day 6 records highest number of clicks in female and male graph. In all the other days, almost equal number of clicks were recorded in female and male respectively. Q2) Dataset Chosen: IPL Dataset(Cricket Dataset) ------------------------------------------------
IPL Dataset (Cricket Dataset) has details of all the matches in all seasons with team1, team2 , toss_winner, toss_decision, winner, win_by_runs, win_by_wickets and other details.
Q2) a. Visualising the number of matches played and victories per team
matches <- read.csv("C:\\Users\\welcome\\Desktop\\Stuff\\MEngg\\Sem-1\\DS\\matches.csv")
library(dplyr)
##
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
##
## filter, lag
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union
matches %>% summarise(match_count = n())
## match_count
## 1 636
library(dplyr)
library(ggplot2)
#wins - groups by winner(team_name) with number of victories
wins = matches %>% group_by(winner) %>% summarise(win_count = n())
ggplot(wins, aes(x=winner,y=win_count)) + geom_bar(fill="#0073C2FF", stat="identity") + geom_text(aes(label=win_count), hjust= -0.3) + coord_flip()
Plot was made to compare between Toss wins and Match wins. How
likely toss winning and match winning are.
toss_win <- matches %>% group_by(toss_winner) %>% summarise(toss_count=n())
wins <- wins [-1,]
total <- data.frame(team=toss_win$toss_winner, tosswin=toss_win$toss_count,matchwin=wins$win_count )
ggplot(total, aes(x=tosswin, y=matchwin, group=1, color=team))+ geom_point(size=3) + geom_line(size=2)
- By removing columns with most number of NA (Made whole column null in the data frame) - Rows having NULL values (or) empty strings are removed.
- Mumbai Indians recorded highest number of victories while kochi tuskers and Pune Warriors record lowest number of victories in all seasons
- Chennai Super Kings, Kolkata Knight Riders toss winning and match winning is most likely followed by Rising Pune Supergaint.