| title: “Cryptography course analysis” |
| author: “Kipsang Mutai Nicholas” |
| date: “01/22/2022” |
| title: “Cryptography course analysis” |
| author: “Kipsang Mutai Nicholas” |
| date: “01/22/2022” |
Online Cryptography analysis
1.Defining the Question.
This research is meant to identify the specific individuals who are likely to click on an advertisement add.
a.) Who is likely to click the online cryptography course advertisement?
2.) Metrics of success
My success will be achieved by correctly identifying individuals that are likely to click into an add. This will be achieved by performing and in depth review of the data at hand to analyse factors influencing the advertisement clicks while assessing individual independent variables and their distribution within the set and their relationship with one another.
3.) The context of analysis.
The online cryptography course is a new field involved with ensuring secure communication between two individuals. This field though new, could improve information security. Using data collected from a former advertisement on a related course posted on a blog, we can get to view the recipient behavior of this advertisements so as to maximize on high priority recipients to ensure effective advertising and return on investment made in the business by not concentrating on low target potential customers.
4.) Experimental design taken
This will involve exhaustive techniques to understand our data in and out. This will be done by finding and dealing with extreem values ,anomalies, missing values and duplicated values to ensure the data used is an actual representative of the actual observations. This will be followed by and exhaustive analysis of the attributes (variables) of the data.Using the inference borrowed from our analysis , we will obtain answers to our specific question while challenging the solution by providing how to make improvements to ensure optimum marketing is achieved.
5.) Dataset appropriateness for analysis
A brief review of the data to inform us what we are working with and its importance for analysis.
# Loading our dataset.
library(data.table)
advert_df<-fread("http://bit.ly/IPAdvertisingData")
# Checking the first 6 observations
head(advert_df)## Daily Time Spent on Site Age Area Income Daily Internet Usage
## 1: 68.95 35 61833.90 256.09
## 2: 80.23 31 68441.85 193.77
## 3: 69.47 26 59785.94 236.50
## 4: 74.15 29 54806.18 245.89
## 5: 68.37 35 73889.99 225.58
## 6: 59.99 23 59761.56 226.74
## Ad Topic Line City Male Country
## 1: Cloned 5thgeneration orchestration Wrightburgh 0 Tunisia
## 2: Monitored national standardization West Jodi 1 Nauru
## 3: Organic bottom-line service-desk Davidton 0 San Marino
## 4: Triple-buffered reciprocal time-frame West Terrifurt 1 Italy
## 5: Robust logistical utilization South Manuel 0 Iceland
## 6: Sharable client-driven software Jamieberg 1 Norway
## Timestamp Clicked on Ad
## 1: 2016-03-27 00:53:11 0
## 2: 2016-04-04 01:39:02 0
## 3: 2016-03-13 20:35:42 0
## 4: 2016-01-10 02:31:19 0
## 5: 2016-06-03 03:36:18 0
## 6: 2016-05-19 14:30:17 0Checking the structure of my dataset.
# Reviewing the dimensions making my table
dimensions <-dim(advert_df)
dimensions## [1] 1000 10My dataset has a thousand observations and 10 variables of numerical,integer character datatypes. There is also time stamp variable within the data.
6.) Cleaning the dataset.
This will be aimed at identifying extreme values, anomalies within the set,missing values and duplicated values in the set.
colnames(advert_df)## [1] "Daily Time Spent on Site" "Age"
## [3] "Area Income" "Daily Internet Usage"
## [5] "Ad Topic Line" "City"
## [7] "Male" "Country"
## [9] "Timestamp" "Clicked on Ad"# Changing the column names for easy readability
colnames(advert_df)<-c("Time_on_site","Age","A_income","Internet_Usage","Ad_Topic","City","Male","Country","Timestamp","Clicked")a.) Numeric datatypes
# Identifying outliers in the numerical columns
library("dplyr") ##
## Attaching package: 'dplyr'## The following objects are masked from 'package:data.table':
##
## between, first, last## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionnums_df<-select_if(advert_df,is.numeric)
nums_df## Time_on_site Age A_income Internet_Usage Male Clicked
## 1: 68.95 35 61833.90 256.09 0 0
## 2: 80.23 31 68441.85 193.77 1 0
## 3: 69.47 26 59785.94 236.50 0 0
## 4: 74.15 29 54806.18 245.89 1 0
## 5: 68.37 35 73889.99 225.58 0 0
## ---
## 996: 72.97 30 71384.57 208.58 1 1
## 997: 51.30 45 67782.17 134.42 1 1
## 998: 51.63 51 42415.72 120.37 1 1
## 999: 55.55 19 41920.79 187.95 0 0
## 1000: 45.01 26 29875.80 178.35 0 1Rechecking the column names
colnames(advert_df)## [1] "Time_on_site" "Age" "A_income" "Internet_Usage"
## [5] "Ad_Topic" "City" "Male" "Country"
## [9] "Timestamp" "Clicked"A function to check for outliers within the numerical values of the set
# Boxplot of the individual variables
box_plot<-function(data,var,main){
boxplot(data[[var]],ylab="Distribution of values",main=main)
}A boxplot of the daily time spent onsite to check for outliers
box_plot(nums_df,1,"A boxplot of Daily Time Spent on Site")
# There are no outliers in the values in the daily time spent on the site A boxplot of Age to check for outliers
box_plot(nums_df,2,"A boxplot of Age")
# There are no outliers in the Age distributionA boxplot of income to check for outliers
box_plot(nums_df,3,"A boxplot of Area Income")
# There are several outliers lying below the 25th percentileA box plot of Daily Internet Usage to check for outliers
box_plot(nums_df,4,"A boxplot of Daily Internet Usage")
# There are no outliers in the daily internet usage valuesA boxplot to check for outliers on whether an individual is male or not
box_plot(nums_df,5,"A boxplot of Males")
# There are no anomalies nor outliers in the male column. Since values are either 0's or 1's for Yes and No respectivelyA boxplot of Clicked on Ad to check for outliers
box_plot(nums_df,6,"A boxplot of Clicked on Ads")
# No anomaly nor outliers detected in this setAnomalies and Outliers in the Non-numeric columns.
Numerics<-noquote(names(nums_df))
Numerics## [1] Time_on_site Age A_income Internet_Usage Male
## [6] Clickednon_nums<-subset(advert_df,select = -c(Time_on_site,Age,A_income,Internet_Usage,Male,Clicked))head(non_nums)## Ad_Topic City Country
## 1: Cloned 5thgeneration orchestration Wrightburgh Tunisia
## 2: Monitored national standardization West Jodi Nauru
## 3: Organic bottom-line service-desk Davidton San Marino
## 4: Triple-buffered reciprocal time-frame West Terrifurt Italy
## 5: Robust logistical utilization South Manuel Iceland
## 6: Sharable client-driven software Jamieberg Norway
## Timestamp
## 1: 2016-03-27 00:53:11
## 2: 2016-04-04 01:39:02
## 3: 2016-03-13 20:35:42
## 4: 2016-01-10 02:31:19
## 5: 2016-06-03 03:36:18
## 6: 2016-05-19 14:30:17Checking for unique values in each column
print(length(unique(non_nums[[1]])))## [1] 1000print(length(unique(non_nums[[2]])))## [1] 969print(length(unique(non_nums[[3]]))) ## [1] 237There are alot of unique values in this set We can see that some cities come in several times as the unique values do not get to the total number of columns There are 237 unique countries in this set. No anomalies detected on an outwardThere are 237 unique countries in this set. No anomalies detected on an outward
Dealing with Outliers in the Area of income set
Checking the numbers outside the InterQuantile range
boxplot.stats(advert_df$A_income)$out## [1] 17709.98 18819.34 15598.29 15879.10 14548.06 13996.50 14775.50 18368.57Checking for legitimacy of the outliers by checking the maximum and the minimum
print(max(advert_df$A_income))## [1] 79484.8print(min(advert_df$A_income))## [1] 13996.5They wont be regarded as illegitimate data points as they are not too extreem
Checking for Duplicates
# Checking for duplicated rows
length(advert_df[duplicated(advert_df),])## [1] 10There are 10 duplicated values in this set ### Dealing with duplicates
clean_df<-advert_df[!duplicated(advert_df),]
clean_df## Time_on_site Age A_income Internet_Usage
## 1: 68.95 35 61833.90 256.09
## 2: 80.23 31 68441.85 193.77
## 3: 69.47 26 59785.94 236.50
## 4: 74.15 29 54806.18 245.89
## 5: 68.37 35 73889.99 225.58
## ---
## 996: 72.97 30 71384.57 208.58
## 997: 51.30 45 67782.17 134.42
## 998: 51.63 51 42415.72 120.37
## 999: 55.55 19 41920.79 187.95
## 1000: 45.01 26 29875.80 178.35
## Ad_Topic City Male
## 1: Cloned 5thgeneration orchestration Wrightburgh 0
## 2: Monitored national standardization West Jodi 1
## 3: Organic bottom-line service-desk Davidton 0
## 4: Triple-buffered reciprocal time-frame West Terrifurt 1
## 5: Robust logistical utilization South Manuel 0
## ---
## 996: Fundamental modular algorithm Duffystad 1
## 997: Grass-roots cohesive monitoring New Darlene 1
## 998: Expanded intangible solution South Jessica 1
## 999: Proactive bandwidth-monitored policy West Steven 0
## 1000: Virtual 5thgeneration emulation Ronniemouth 0
## Country Timestamp Clicked
## 1: Tunisia 2016-03-27 00:53:11 0
## 2: Nauru 2016-04-04 01:39:02 0
## 3: San Marino 2016-03-13 20:35:42 0
## 4: Italy 2016-01-10 02:31:19 0
## 5: Iceland 2016-06-03 03:36:18 0
## ---
## 996: Lebanon 2016-02-11 21:49:00 1
## 997: Bosnia and Herzegovina 2016-04-22 02:07:01 1
## 998: Mongolia 2016-02-01 17:24:57 1
## 999: Guatemala 2016-03-24 02:35:54 0
## 1000: Brazil 2016-06-03 21:43:21 1confirming the operation
clean_df[duplicated(clean_df),]## Empty data.table (0 rows and 10 cols): Time_on_site,Age,A_income,Internet_Usage,Ad_Topic,City...Checking for null values
# Length which is null
length(which(is.na.data.frame(clean_df)))## [1] 0There are no missing values in the set
Reextracting numerical variables from clean data
# Re-extracting numerical values from the clean data with no duplicates
library("dplyr")
clean_nums_df<-select_if(clean_df,is.numeric)
clean_nums_df## Time_on_site Age A_income Internet_Usage Male Clicked
## 1: 68.95 35 61833.90 256.09 0 0
## 2: 80.23 31 68441.85 193.77 1 0
## 3: 69.47 26 59785.94 236.50 0 0
## 4: 74.15 29 54806.18 245.89 1 0
## 5: 68.37 35 73889.99 225.58 0 0
## ---
## 996: 72.97 30 71384.57 208.58 1 1
## 997: 51.30 45 67782.17 134.42 1 1
## 998: 51.63 51 42415.72 120.37 1 1
## 999: 55.55 19 41920.79 187.95 0 0
## 1000: 45.01 26 29875.80 178.35 0 17. Univariate Analysis
Measures of Central Tendacy
# Numerical values
head(clean_nums_df)## Time_on_site Age A_income Internet_Usage Male Clicked
## 1: 68.95 35 61833.90 256.09 0 0
## 2: 80.23 31 68441.85 193.77 1 0
## 3: 69.47 26 59785.94 236.50 0 0
## 4: 74.15 29 54806.18 245.89 1 0
## 5: 68.37 35 73889.99 225.58 0 0
## 6: 59.99 23 59761.56 226.74 1 0Mean of individual columns
# Mean of individual columns
colMeans(clean_nums_df)## Time_on_site Age A_income Internet_Usage Male
## 65.0002 36.0090 55000.0001 180.0001 0.4810
## Clicked
## 0.5000The values as valid especially for the 0’s and 1’s columns the mean is approximately 0.5 giving us a clue of the distribution of the discrete values.
#A summary of everything in numerical set
summary.data.frame(clean_nums_df)## Time_on_site Age A_income Internet_Usage
## Min. :32.60 Min. :19.00 Min. :13996 Min. :104.8
## 1st Qu.:51.36 1st Qu.:29.00 1st Qu.:47032 1st Qu.:138.8
## Median :68.22 Median :35.00 Median :57012 Median :183.1
## Mean :65.00 Mean :36.01 Mean :55000 Mean :180.0
## 3rd Qu.:78.55 3rd Qu.:42.00 3rd Qu.:65471 3rd Qu.:218.8
## Max. :91.43 Max. :61.00 Max. :79485 Max. :270.0
## Male Clicked
## Min. :0.000 Min. :0.0
## 1st Qu.:0.000 1st Qu.:0.0
## Median :0.000 Median :0.5
## Mean :0.481 Mean :0.5
## 3rd Qu.:1.000 3rd Qu.:1.0
## Max. :1.000 Max. :1.0Histogram of the individual columns to see their distribution
For continous variables a histogram is paramount to display the distribution of values in the set while telling us the skewness and kurtosis of out data
hist(clean_nums_df$Age,xlab="Age",main="Distribution of age")
We can see from this that age of participants was mostly concentrated around 30 years. Its skewed to the left giving us insight that most participants in this sample set were in the mid 30’s
Distribution of the time on sight
hist(clean_nums_df$Time_on_site,xlab = "Time spent on site",main="The distribution of the time on site")
Frequency of Time spent is relatively high the 3rd quantile of the set. This tells us that a significantly large amount of time is spent on the site
# Distribution of area income
hist(clean_nums_df$A_income,xlab="Area Income",main="This is the ditribution of the Area Income")
The Area income is skewed to the right, the tells us that the income is populated on the higher end with a low population with lower income
# Internet usage
hist(clean_nums_df$Internet_Usage,xlab="Internet Usage",main="Distribution of Internet usage")
The distribution is relatively normal with the lower population on the higher end.
Discrete Variables
# Getting the values in the male variable
males<-clean_nums_df$Male
# Getting the frequency table of the male set
male_dist<-table(males)
# Plotting a bar plot to understand the distribution of discrete values in the males column
barplot(male_dist,main = "The Ditribution of Males",xlab = "Males and Non males")
# The non male which i would presume to be female were higher in count as compared to males
# Getting the values in the clicks
ad_clicks<-clean_nums_df$Clicked
# Getting the frequency table of the clicks set
ad_clicks_dist<-table(ad_clicks)
# Plotting a bar plot to understand the distribution of discrete values in the clicks column
barplot(ad_clicks_dist,main = "The Ditribution of Adclicks",xlab = "Yes and No")
The distribution was balanced.
8.) Bivariate and Multivariate Analysis
Looking at the relationship between two variables and their variations within the set.
install.packages("Hmisc")## Installing package into 'C:/Users/RoySambu/Documents/R/win-library/4.0'
## (as 'lib' is unspecified)## package 'Hmisc' successfully unpacked and MD5 sums checked
##
## The downloaded binary packages are in
## C:\Users\RoySambu\AppData\Local\Temp\Rtmp82YnjW\downloaded_packageslibrary("Hmisc")## Loading required package: lattice## Loading required package: survival## Loading required package: Formula## Loading required package: ggplot2##
## Attaching package: 'Hmisc'## The following objects are masked from 'package:dplyr':
##
## src, summarize## The following objects are masked from 'package:base':
##
## format.pval, units# Checking the correlation and their corresponding significant levels
correlations<-rcorr(as.matrix(clean_nums_df))
correlations## Time_on_site Age A_income Internet_Usage Male Clicked
## Time_on_site 1.00 -0.33 0.31 0.52 -0.02 -0.75
## Age -0.33 1.00 -0.18 -0.37 -0.02 0.49
## A_income 0.31 -0.18 1.00 0.34 0.00 -0.48
## Internet_Usage 0.52 -0.37 0.34 1.00 0.03 -0.79
## Male -0.02 -0.02 0.00 0.03 1.00 -0.04
## Clicked -0.75 0.49 -0.48 -0.79 -0.04 1.00
##
## n= 1000
##
##
## P
## Time_on_site Age A_income Internet_Usage Male Clicked
## Time_on_site 0.0000 0.0000 0.0000 0.5495 0.0000
## Age 0.0000 0.0000 0.0000 0.5062 0.0000
## A_income 0.0000 0.0000 0.0000 0.9667 0.0000
## Internet_Usage 0.0000 0.0000 0.0000 0.3762 0.0000
## Male 0.5495 0.5062 0.9667 0.3762 0.2296
## Clicked 0.0000 0.0000 0.0000 0.0000 0.2296There is a negative linear correlation between Age and Time on site. This means for any positive variation in Age there is a negative variation of 0.33 magnitude. It is true because it is rare to find people who are advanced in age spending time on the internet.This can also be reflected in internet usage negative correlation with age.
There is a positive linear relationship between Area income and time on site.
The P-values explain the probability if the correlation being due to chance or an equally extreeme event. We get to see the correlations that are statistically significant being (P<0.05):
All except Male and the other continuous variables - This is the case because male has categorical or rather discrete values thus tabulation if its correlation with a continuous variable is not statistically significant.
The same sentiment would be shared by the clicked variable. The correlations shown with other variables would not be a representation of the numbers on the ground as clicked is discrete and the values are either 0 or 1. We cannot use correlation to make a conclusion about the relationship between clicked and other continous variables.
Visualizations
# PLotting a box plot to assess the relationship betwen Age of individuals and their click rate.
boxplot(Age~Clicked,data=clean_df,main = "The Ditribution of Ages and clicks",xlab = "No and Yes")
# It is evident that individuals in between the age of approximately 35 to 45 are likely to click into the add as compared to those in their 20's to 30's and the age above 50.Plotting relationships between clicked and income across areas.
# Income area and the clicks
boxplot(A_income~Clicked,data=clean_df,main = "The Ditribution of Area Income area and clicks",xlab = "No and Yes")
# It is so evident that the people in areas with income between 40,000 and approximately 59,000 are more likely to click into an add as compared with individuals with areas of and income above 59,000 to approximately 70,000. some individuals with and income below 40,000 are seem NOT to click into an add. Relationship between time on site and clicks.
# Plot to understand relationship between time on side and ad clicks
boxplot(Time_on_site~Clicked,data=clean_df,main = "The Ditribution of Time onsite and clicks",xlab = "No and Yes")
### This is an interesting discovery. People who spend less time on the blogging site are more likely to click into an add as compared to people who spend more time on the sight. This could be in avoidance of distraction or just by lack of interest.
## A hypothesis testing on this needs to be made to ascertain the correctness on this discovery.
## A few individuals who spend less time on the site(in units specified) do not also click into the add. It is safe to say that the target individuals are those who spend from 45 to 60 (units of time specified).Relationship between Internet Usage and Clicks
# Internet usage and clicks
boxplot(Internet_Usage~Clicked,data=clean_df,main = "The Ditribution of Internet usage and clicks",xlab = "No and Yes")
# Same as time on site, people with less internet usage are more likely to click into an add. There are however a few individuals who have a higher internet usage that still click into the add. The least likely individuals to click into an add use internet(units specified) between 200 and approximately 230.Finding the relationship between the gender and the ads clicked.
# Finding the relationship between male and clicked
my_table<-with(clean_df,table(Clicked,Male))
print(my_table)## Male
## Clicked 0 1
## 0 250 250
## 1 269 231# From this we can see that the NON-MALE are more likely to click as compared to Male. For those who did not click, the ratio is the same. Although its important to say as we had seen in the correlation this relationship is not statistically significant and thus should not be considered as a factor for our target audience in this case.Discrete variables
# Dealing with yes clicks to see the countries that are most frequent
yes_clicked<-clean_df[clean_df$Clicked==1]
# Checking the frequency of individual countries
country_tbl<-as.matrix(table(yes_clicked$Country))
country_tbl[country_tbl>=5,]## Afghanistan Australia Ethiopia France Hungary
## 5 7 7 5 5
## Liberia Liechtenstein Mayotte Peru Senegal
## 6 6 5 5 5
## South Africa Turkey
## 6 7# Its evident that Australia,France,Hungary, Turkey and other countries with a frequency count greater than five can be a priority while displaying adsunique(country_tbl)## [,1]
## Afghanistan 5
## Albania 4
## Algeria 3
## Andorra 2
## Angola 1
## Australia 7
## Liberia 6City
# Dealing with yes clicks to see the cities that are most frequent
yes_clicked<-clean_df[clean_df$Clicked==1]Checking the frequency of individual countries
city_tbl<-as.matrix(table(yes_clicked$City))
unique(city_tbl)## [,1]
## Adamsbury 1
## Lake David 2# Lake David appears uniquely twice in the cities with ad clicks9.) Findings and Recomendations
Based on the findings from our data we get to see that.
- Non-Males are more likely to click into an add as compared to males.
- Countries that have a higher count include but are not limited to Australia,Turkey,France, Hungary etc.
- Internet usage is inversely proportional to add clicks . This can however be challenged by checking the statistical significance of that finding.
- Ad clicks are also inversely proportional to time spent on site. I can challenge this by performing a hypothesis testing on this findings to prove its statistical significance.
- The age that is more likely to click an ad is between approximately 35 and 45 yrs.
- The area income that is likely to click in an add fall between approximately 40,000 and 59,000 .
- This can be challenged by looking deeper into these groups and assessing whether the extent of influence our outliers regarded as legitimate data points had on the analysis.
10.) Conclusions and Recommendations
Based on our findings, we get to conclude that:
- The location, Income ,internet usage and time spent on the blogging site is a factor to consider in considering high priority customers.
- More data needs to be collected on locations to understand the populations distribution in these areas.
- Gender is not a cardinal factor to consider as its relationship with ads clicked is not statistically significant its probability of being due to chance surpasses our thresh hold of 0.05, which is our margin of error.