R code to generate percentage of BJ DNS traffic using Google PDNS (8.8.8.8)

bj_dns_traffic_to_google_pdns.r

# Get data from ApNic
# https://stats.labs.apnic.net/ecdsa/BJ
# For instance
cat bj_gpdns.csv
ISP,Sample,GPDNS %
CanalBox-Benin-AS,1299,61.12
SUD-TELCOM-AS,186,34.41
BENINTELECOM,35118,92.57
Spacetel,60193,70.54
JENY-SAS-AS,3091,55.71
OTI-AS,3046,19.21
ISOCEL,13149,15.27
ETISALAT-AS,31734,44.57
UNIVERCELL-AS,1977,1.67

# Get dataframe
pdns <- read.csv('bj_gpdns.csv', header=TRUE)

#Create new column to store number of sample data going to GPDNS per ISP
pdns$Sample_GPDNS <- (pdns$Sample * pdns$GPDNS) / 100

#Compute global percentage of sample data going to GPDNS per ISP
pdns$Global_GPDNS <- (pdns$Sample_GPDNS / sum(pdns$Sample) )* 100

#Get percentage of sample data going to GPDNS
sum(pdns$Sample_GPDNS) / sum(pdns$Sample) * 100
[1] 62.96601

#Get per ISP, percentage of sample going to GPDNS
(pdns$Sample_GPDNS / sum(pdns$Sample) )* 100
[1]  0.53003064  0.04272736 21.70243776 28.34587878  1.14958383  0.39063014  1.34041798  9.44225952
[9]  0.02204102

#Verify that wa have the same percentage
(sum(pdns$Sample_GPDNS) / sum(pdns$Sample) * 100) == sum((pdns$Sample_GPDNS / sum(pdns$Sample) )* 100)


#Plot
#ggplot(pdns, aes(x = ISP, y = Global_GPDNS)) + ylab("% to GPDN") + geom_col(aes(x = ISP, fill = factor(ISP)))

#ggplot(pdns, aes(x=ISP,y=Global_GPDNS,fill=ISP))  + geom_col() + ylab("% to GPDN")
#ggplot(pdns, aes(x=ISP,y=Global_GPDNS,fill=ISP))  + geom_col() + ylab("% to GPDN") + guides(fill=FALSE)
#ggplot(pdns, aes(x=ISP,y=Global_GPDNS,fill=ISP))  + geom_bar(stat="identity")  + ylab("% to GPDN") + guides(fill=FALSE)
ggplot(pdns, aes(x=reorder(ISP, -Global_GPDNS),y=Global_GPDNS,fill=ISP))  + geom_bar(stat="identity")  + ylab("% to GPDN") + guides(fill=FALSE) + xlab("ISPs")
#ggplot(pdns, aes(x=reorder(ISP, -Global_GPDNS),y=Global_GPDNS,fill=ISP, label=Global_GPDNS))  + geom_col()  + ylab("% to GPDNS") + guides(fill=FALSE) + xlab("ISPs") + theme() + geom_text(vjust=-1)