AIM

This is the part 2 of this project.This is a data analytics project for mining analyzing, visualizing the data collected by the Kaggle Data science survey conducted in 2017.

1) Part 2- Will mostly aim at analyzing things such as what is the most preferred tool for implementing Datascience, Most used ML method, learning platform,What language they recommend etc and other more specific answers related to datascience and machine learning given by the survey participants.

Let’s get started

Let’s start with the most preferred tool for implementing DS and ML.

require(data.table)
require(highcharter)
require(ggplot2)
require(tidyverse)

SurveyDf<-fread("../Datasets/kagglesurvey2017/multipleChoiceResponses.csv") #for faster data reading

attach(SurveyDf)

table(MLToolNextYearSelect)
## MLToolNextYearSelect
##                                                    
##                                               5718 
##                            Amazon Machine Learning 
##                                                392 
##                                Amazon Web services 
##                                                273 
##                                             Angoss 
##                                                  1 
##                                              C/C++ 
##                                                186 
##                                           Cloudera 
##                                                 44 
##                                          DataRobot 
##                                                220 
##                                              Flume 
##                                                 16 
##                               Google Cloud Compute 
##                                                296 
##                                    Hadoop/Hive/Pig 
##                                                417 
##     I don't plan on learning a new tool/technology 
##                                                341 
##                                         IBM Cognos 
##                                                  7 
##                                   IBM SPSS Modeler 
##                                                 19 
##                                IBM SPSS Statistics 
##                                                 24 
##                       IBM Watson / Waton Analytics 
##                                                194 
##                                             Impala 
##                                                  9 
##                                               Java 
##                                                116 
##                                              Julia 
##                                                222 
##                                  Jupyter notebooks 
##                                                358 
##                         KNIME (commercial version) 
##                                                 14 
##                               KNIME (free version) 
##                                                 38 
##                                        Mathematica 
##                                                 64 
##                                      MATLAB/Octave 
##                                                115 
##                   Microsoft Azure Machine Learning 
##                                                220 
##                        Microsoft Excel Data Mining 
##                                                 46 
## Microsoft R Server (Formerly Revolution Analytics) 
##                                                 65 
##                   Microsoft SQL Server Data Mining 
##                                                 31 
##                                            Minitab 
##                                                  8 
##                                              NoSQL 
##                                                 94 
##            Oracle Data Mining/ Oracle R Enterprise 
##                                                 25 
##                                             Orange 
##                                                 15 
##                                              Other 
##                                                407 
##                                               Perl 
##                                                 13 
##                                             Python 
##                                               1713 
##                                           QlikView 
##                                                 20 
##                                                  R 
##                                                910 
##                    RapidMiner (commercial version) 
##                                                 18 
##                          RapidMiner (free version) 
##                                                 36 
##           Salfrod Systems CART/MARS/TreeNet/RF/SPM 
##                                                 11 
##           SAP BusinessObjects Predictive Analytics 
##                                                 81 
##                                           SAS Base 
##                                                 57 
##                               SAS Enterprise Miner 
##                                                 50 
##                                            SAS JMP 
##                                                  7 
##                                      Spark / MLlib 
##                                                755 
##                                                SQL 
##                                                138 
##                                               Stan 
##                                                 64 
##          Statistica (Quest/Dell-formerly Statsoft) 
##                                                 15 
##                                            Tableau 
##                                                150 
##                                         TensorFlow 
##                                               2621 
##                                     TIBCO Spotfire 
##                                                  6 
##                                   Unix shell / awk 
##                                                 25 
##                                               Weka 
##                                                 31
tooldf<-as.data.frame(table(MLToolNextYearSelect)) %>% arrange(desc(Freq))
#let's remove missing value
tooldf[1,]<-NA
tooldf<-na.omit(tooldf)
names(tooldf)<-c("Tool","Count")
#now let's plot the data
hchart(tooldf,hcaes(x=Tool,y=Count),type="column",name="Count",color="#9B6ED8") %>%  
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Barplot of tools used by participants",align="center") %>%
  hc_add_theme(hc_theme_elementary()) 
toptooldf<-na.omit(tooldf) %>% arrange(desc(Count)) %>% top_n(10)
#plotting a funnel chart of top 10 most used tools entered by the users
col <- c("#d35400", "#2980b9", "#2ecc71", "#f1c40f", "#2c3e50", "#7f8c8d","#000004", "#3B0F70", "#8C2981", "#DE4968")

hchart(toptooldf,hcaes(x=Tool,y=Count),type="funnel",name="Count",color=col) %>% 
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Funnel chart of top 10 most used tools used by survey participants",align="center")

Let’s check Top countries uses which tools ?

Let’s first make a dataframe which consists of Top 10 particapnt’s countries and grouped by the preferred tool entered by them which is then summarized by the total count.

#dataframe of top 10 countries 
countryCount<-as.data.frame(table(SurveyDf$Country)) %>%  top_n(10)


countryTool<-SurveyDf %>% group_by(MLToolNextYearSelect,Country) %>%
  select(Country,MLToolNextYearSelect) %>% 
  filter(Country %in% countryCount$Var1, MLToolNextYearSelect %in% toptooldf$Tool) %>%
  summarise(total_count=n()) %>%
  arrange(desc(total_count))


hchart(countryTool,hcaes(x=Country,y=total_count,group=MLToolNextYearSelect),type="column") %>% 
  hc_title(text="Barplot of Top 10 Country grouped by Tool",align="center") %>%
  hc_exporting(enabled=TRUE) 
ggplot(aes(x=Country,y=total_count),data=countryTool) +
  geom_col(fill="purple") +
  coord_flip() +
  facet_wrap(~MLToolNextYearSelect)

We can notice that Every top 10 country uses Tensorflow, followed by python , then R as their preferred tools for implementing machine learning and data science.

Let’s see which gender prefers which tool amongst the top 10 most used tools?

genderTool<-SurveyDf %>% group_by(MLToolNextYearSelect,GenderSelect) %>%
  select(GenderSelect,MLToolNextYearSelect) %>% 
  filter(MLToolNextYearSelect %in% toptooldf$Tool,GenderSelect %in% c("Male","Female")) %>%
  summarise(total_count=n()) %>%
  arrange(desc(total_count))


hchart(genderTool,hcaes(x=MLToolNextYearSelect,y=total_count,group=GenderSelect),type="column") %>%   hc_title(text="Gender vs Most preferred tool",align="center") %>%
  hc_exporting(enabled=TRUE) 
#checking the ratio of male to female users for tools-
genderTooldf<-genderTool %>% spread(key = GenderSelect,value=total_count) %>%
  mutate(Percent_male=round((Male/(Male+Female))*100,2),Percent_female=round((Female/(Male+Female))*100,2)) %>%
  arrange()

hchart(genderTooldf,hcaes(x=MLToolNextYearSelect,y=Percent_female),name="Percent",type="column",color="#ED2FAE") %>%   hc_title(text="Percent of female using tools out of 100",align="center") %>%
  hc_exporting(enabled=TRUE) 
hchart(genderTooldf,hcaes(x=MLToolNextYearSelect,y=Percent_male),name="Percent",type="column",color="#F36A20") %>%   hc_title(text="Percent of Male using tools out of 100",align="center") %>%
  hc_exporting(enabled=TRUE)

Now let’s see which ages group of people use which tool. I will only check the distribution of the top used tools and the ages of participants.

AgeTooldf<-SurveyDf %>% group_by(MLToolNextYearSelect) %>%
  filter(MLToolNextYearSelect %in% c("TensorFlow","R","Python","Spark / MLlib","Amazon Web services","Hadoop/Hive/Pig","Google Cloud Compute","Jupyter notebooks")) %>% 
  select(MLToolNextYearSelect,Age)

hcboxplot(x = AgeTooldf$Age, var = AgeTooldf$MLToolNextYearSelect ,name = "Age", color = "#FF5733",outliers = FALSE) %>%
  hc_chart(type="column") %>% 
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Boxplot of Top 3 Ml tools used by participants and their Ages",align="center") %>% 
  hc_add_theme(hc_theme_elementary())

From the above plots R has highest median age of 31 , compared to other tools. Whereas Hadoop and Python are used by relatively younger data scientists as inferred from this boxplot.

Let’s check the mean ages of the participants and the tools entered and used by them and plot them. Let’s summarize the data using dplyr package.

meanAgeTool<-SurveyDf %>% group_by(MLToolNextYearSelect) %>%
 summarise(meanAge=mean(Age,na.rm=T)) %>%      
  arrange(desc(meanAge))

#rounding off the mean ages
meanAgeTool$meanAge<-round(meanAgeTool$meanAge,1)
names(meanAgeTool)<-c("Tool","MeanAge")#renaming the columns


topAgeTool<-meanAgeTool %>% filter(Tool %in% toptooldf$Tool)
col<-c("#d35400", "#2980b9", "#2ecc71", "#f1c40f", "#2c3e50", "#7f8c8d","#000004", "#3B0F70", "#8C2981", "#DE4968")
#let's make a scatterplot

hchart(meanAgeTool,hcaes(x=Tool,y=MeanAge),name="Tool" ,type="scatter",color="#F44E1D") %>% 
  hc_chart(type="column") %>% 
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Scatter plot of ML tools used by participants and their Mean  
           Ages",align="center") %>% 
  hc_add_theme(hc_theme_elementary())
hchart(topAgeTool,hcaes(x=Tool,y=MeanAge,size=MeanAge),color="#7CE3B0",name="Mean Age" ,type="bubble") %>% 
  hc_chart(type="column") %>% 
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Bubble chart of top 10 ML tools used by participants and their Mean Ages",align="center") %>% 
  hc_add_theme(hc_theme_elementary())

From the above plot of mean ages , we can see that R users have higher mean age than Python,Tensorflow and Hadoop users.This means Python and Hadoop is more famous and used amongst relatively younger data scientists.

Let’s analyze the most used ML method?

table(MLMethodNextYearSelect)
## MLMethodNextYearSelect
##                                             
##                                        5883 
##                           Anomaly Detection 
##                                         307 
##                           Association Rules 
##                                          46 
##                            Bayesian Methods 
##                                         511 
##                            Cluster Analysis 
##                                         216 
##                              Decision Trees 
##                                         156 
##                               Deep learning 
##                                        4362 
##   Ensemble Methods (e.g. boosting, bagging) 
##                                         269 
##                             Factor Analysis 
##                                          76 
##           Genetic & Evolutionary Algorithms 
##                                         425 
## I don't plan on learning a new ML/DS method 
##                                         195 
##                               Link Analysis 
##                                          55 
##                                        MARS 
##                                          11 
##                         Monte Carlo Methods 
##                                         230 
##                                 Neural Nets 
##                                        1386 
##                                       Other 
##                                         258 
##                      Proprietary Algorithms 
##                                          98 
##                              Random Forests 
##                                         154 
##                                  Regression 
##                                         220 
##                              Rule Induction 
##                                          41 
##                     Social Network Analysis 
##                                         364 
##               Support Vector Machines (SVM) 
##                                         147 
##                           Survival Analysis 
##                                         104 
##                                 Text Mining 
##                                         493 
##                        Time Series Analysis 
##                                         680 
##                             Uplift Modeling 
##                                          29
Mlmethod<-as.data.frame(table(MLMethodNextYearSelect)) %>% arrange(desc(Freq))

Mlmethod[1,]<-NA
names(Mlmethod)<-c("Method","Count")

hchart(na.omit(Mlmethod),hcaes(x=Method,y=Count),type="column",name="Count",color="#FF5733") %>% 
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Barplot of ML Method used by participants",align="center") %>%
  hc_add_theme(hc_theme_elementary()) 
#treemap of most preferred methods of ML
hchart(Mlmethod, "treemap", hcaes(x = Method, value = Count,color=Count)) %>%
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Treemap of ML Method used by participants",align="center")

As we can notice the most used ML technique is Deep learning and neural networks followed by time series analysis.

Let’s check the distribution of ages of participants and their preferred method.

hcboxplot(x = SurveyDf$Age, var =SurveyDf$MLMethodNextYearSelect ,name = "Age", color = "#E127AB",outliers = FALSE) %>%
  hc_chart(type="column") %>% 
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Boxplot of ML Method used by participants and their Ages",align="center") %>% 
  hc_add_theme(hc_theme_elementary())

Let’s check the countrywise usage of ML methods.Again I am going to do it for top 10 countries.

countryMethod<-SurveyDf %>% select(Country,MLMethodNextYearSelect) %>%
  group_by(Country,MLMethodNextYearSelect) %>%
  filter(Country %in% countryCount$Var1) %>%
  summarize(total_count=n())%>%
  arrange(desc(total_count))

Let’s first filter for United States and check the most used tools.

#First for United states
USmethoddf<-countryMethod %>% filter(Country=="United States") %>% arrange(desc(total_count))

USmethoddf[1,]<-NA

hchart(na.omit(USmethoddf),hcaes(x=MLMethodNextYearSelect,y=total_count),type="column",color="#BFD13D") %>%
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Barplot of ML Method used by participants in United States",align="center") %>%
  hc_add_theme(hc_theme_elementary())

For India.

Indiamethoddf<-countryMethod %>% filter(Country=="India") %>% arrange(desc(total_count))

Indiamethoddf[1,]<-NA

hchart(na.omit(Indiamethoddf),hcaes(x=MLMethodNextYearSelect,y=total_count),type="column",color="red") %>%
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Barplot of ML Method used by participants in India",align="center") %>%
  hc_add_theme(hc_theme_elementary())

For China.

ChinaMethoddf<-countryMethod %>% filter(Country=="People 's Republic of China") %>% arrange(desc(total_count))

ChinaMethoddf[2,]<-NA

hchart(na.omit(ChinaMethoddf),hcaes(x=MLMethodNextYearSelect,y=total_count,color=MLMethodNextYearSelect),type="column") %>%
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Barplot of ML Method used by participants in China",align="center") %>%
  hc_add_theme(hc_theme_elementary())

For Russia.

RussiaMethoddf<-countryMethod %>% filter(Country=="Russia") %>% arrange(desc(total_count))

RussiaMethoddf[1,]<-NA

hchart(na.omit(RussiaMethoddf),hcaes(x=MLMethodNextYearSelect,y=total_count,color=MLMethodNextYearSelect),type="column") %>%
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Barplot of ML Method used by participants in Russia",align="center") %>%
  hc_add_theme(hc_theme_elementary())

For United Kingdom.

UKMethoddf<-countryMethod %>% filter(Country=="United Kingdom") %>% arrange(desc(total_count))

UKMethoddf[1,]<-NA

hchart(na.omit(UKMethoddf),hcaes(x=MLMethodNextYearSelect,y=total_count,color=MLMethodNextYearSelect),type="column") %>%
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Barplot of ML Method used by participants in UK",align="center") %>%
  hc_add_theme(hc_theme_elementary())

USA,India and China combined.

MethodTopCountry<-countryMethod %>% filter(Country %in% c("United States","India","United Kingdom"))

MethodTopCountry[c(1,2,7),]<-NA


hchart(na.omit(MethodTopCountry),hcaes(x=MLMethodNextYearSelect,y=total_count,group=Country),type="column",color=c("#FFA500","#C71585","#00FF00")) %>%
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Barplot of ML Method used by top 3 participants Countries",align="center") %>%
  hc_add_theme(hc_theme_elementary())

Let’s check from where participants collect data?

Let’s find the top 20 sources from where participants collect data. This is a multiple choice field. So it has multiple answers.

publicData<-as.data.frame(table(PublicDatasetsSelect)) %>% arrange(desc(Freq)) %>% top_n(20)
publicData[1,]<-NA
names(publicData)<-c("source",'count')

hchart(na.omit(publicData),hcaes(x=source,y=count),type="column",color="#26BA0B")

So we discover that mostly people prefer online platforms such as Kaggle, data.worlds etc to download public data sets, followed by google search and scraping data from web and github as data sources to do data science and analytics projects.

Let’s check which is the favourite Platform to learn Data science?

The filed which I am going to use is a multiple choice field which a survey participant can select multiple answers.

#Let's check for the top 10 learning platforms
platformdf<-SurveyDf %>% group_by(LearningPlatformSelect) %>%
  summarise(count=n()) %>% top_n(20) %>% arrange(desc(count))
platformdf[1,]<-NA

hchart(na.omit(platformdf),hcaes(x=LearningPlatformSelect,y=count),type="column",color="yellow") %>% 
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Barplot of top 20 platforms used by participants to learn Data science",align="center") %>% 
  hc_add_theme(hc_theme_elementary())

As expected most people prefer Kaggle, Online MOOC courses to learn data science related topics and subjects.

Now let’s check and do some bi-variate analysis of Learning platform with other major variables.

  1. First let’s check which age group prefers which platform. Let’s do some data aggregation and summarization to find mean of ages grouped by the learning platform.
ageplatform <-SurveyDf %>% filter(LearningPlatformSelect %in% platformdf$LearningPlatformSelect) %>% 
  group_by(LearningPlatformSelect) %>% 
  select(Age,LearningPlatformSelect)

ageplatform<-na.omit(ageplatform)

#let's first make a boxplot

hcboxplot(x = ageplatform$Age , var = ageplatform$LearningPlatformSelect,name = "Age", color = "cayn",outliers = FALSE) %>%
  hc_chart(type="column") %>% 
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Boxplot of learning platform used by participants and their Ages",align="center") %>% 
  hc_add_theme(hc_theme_elementary()) 
by(ageplatform$Age,ageplatform$LearningPlatformSelect,summary) #summary statistics of age grouped by learning platform
## ageplatform$LearningPlatformSelect: Arxiv
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   18.00   24.00   26.00   30.11   35.00   66.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Blogs
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   21.00   26.00   29.00   31.91   35.00   65.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: College/University
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    17.0    23.0    27.0    30.3    32.5   100.0 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Kaggle
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   17.00   24.00   29.00   31.76   36.00   99.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Kaggle,Online courses
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   19.00   25.00   30.00   31.56   37.00   65.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Kaggle,Online courses,Personal Projects,Stack Overflow Q&A
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   17.00   25.00   31.50   34.11   38.00   60.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Kaggle,Online courses,Stack Overflow Q&A
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   19.00   26.50   32.00   34.46   41.75   59.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Kaggle,Online courses,Stack Overflow Q&A,YouTube Videos
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   19.00   25.00   28.50   30.37   35.00   48.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Kaggle,Online courses,YouTube Videos
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   18.00   26.25   33.00   35.26   42.75   61.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Kaggle,Personal Projects
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   17.00   22.50   32.00   33.33   41.50   59.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Kaggle,Stack Overflow Q&A
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   18.00   26.00   31.50   33.05   38.25   58.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Kaggle,Stack Overflow Q&A,YouTube Videos
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   14.00   23.00   27.00   30.21   35.75   56.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Kaggle,YouTube Videos
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   20.00   25.00   29.00   30.39   34.00   53.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Online courses
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##     0.0    25.0    30.0    32.4    37.0    75.0 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Online courses,YouTube Videos
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   17.00   23.00   29.00   32.06   38.00   63.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Other
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   21.00   28.00   33.00   36.49   42.00   80.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Personal Projects
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   17.00   24.00   29.00   33.92   39.00   99.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Stack Overflow Q&A
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   16.00   25.00   28.00   30.92   35.00   64.00 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: Textbook
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    19.0    25.5    31.0    34.4    39.0    77.0 
## -------------------------------------------------------- 
## ageplatform$LearningPlatformSelect: YouTube Videos
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   16.00   23.00   26.00   30.34   35.00   63.00

From the above boxplot we can see that younger participants generally prefer arvix and youtube videos more as the median ages for these platforms is least i.e age of 26.

Let’s summarize and calculate mean ages for more clear picture and understanding.

MeanAgePlatform<-ageplatform %>% group_by(LearningPlatformSelect) %>%
  summarise(mean_age=round(mean(Age),1)) %>% arrange(desc(mean_age))


hchart(MeanAgePlatform,hcaes(x=LearningPlatformSelect,y=mean_age),type="scatter",name="Mean Age",color="#596536") %>% 
  hc_exporting(enabled = TRUE) %>%
  hc_title(text="Scatter plot of mean of ages and the learning platform",align="center") %>% 
  hc_add_theme(hc_theme_elementary())

Pie charts of the usefullness of each of the learning platform entered by the survey participants.

df<- SurveyDf %>% select(LearningPlatformUsefulnessArxiv,LearningPlatformUsefulnessYouTube,LearningPlatformUsefulnessKaggle,LearningPlatformUsefulnessCollege,LearningPlatformUsefulnessCourses) 
df<-df[complete.cases(df),]



  # Data
  hchart(df$LearningPlatformUsefulnessArxiv, "pie") %>% 
    hc_exporting(enabled = TRUE) %>%
    hc_title(text="Pie chart of Question How usefull is Arxiv as learning platform",align="center") %>% 
  hc_add_theme(hc_theme_elementary())
  hchart(df$LearningPlatformUsefulnessCourses, "pie")  %>% 
    hc_exporting(enabled = TRUE) %>%
  hc_title(text="Pie chart of Question How usefull are Online courses as learning platform",align="center") %>% 
  hc_add_theme(hc_theme_elementary()) 
  hchart(df$LearningPlatformUsefulnessCollege, "pie")  %>% 
    hc_exporting(enabled = TRUE) %>%
  hc_title(text="Pie chart of Question How usefull us Collage as learning platform",align="center") %>% 
  hc_add_theme(hc_theme_elementary()) 
  hchart(df$LearningPlatformUsefulnessKaggle, "pie")  %>% 
    hc_exporting(enabled = TRUE) %>%
  hc_title(text="Pie chart of Question How usefull is Kaggle as learning platform",align="center") %>% 
  hc_add_theme(hc_theme_elementary()) 
  hchart(df$LearningPlatformUsefulnessYouTube, "pie") %>%  
    hc_exporting(enabled = TRUE) %>%
  hc_title(text="Pie chart of Question How usefull is Youtube as learning platform",align="center") %>% 
  hc_add_theme(hc_theme_elementary())