Anly 506 - Exploratory Data Analytics Final Project Report

Research Question

What are top three regions with highest GDP per capita from year 1800 to 2015 ?
How is the population growth over years for each region ?
Which all regions is close to overall average Life expectancy?
What are most ppoulated South Asian Countries since 1900 ? How is the growth, is it exponential ?
Considering Life Expectancy, among America & South Asia region which region has more GDP per Capita?
How Region’s GDP per Capita (in $) coorelate with Life Expectancy?

Data Description

Overall narrative summary of data:
- There are total 41284 observation in our gapminder dataset.
- There are total 6 variable in our dataset.
- Below is the Data type of variables
  1. Country - Character
  2. Year - Numeric
  3. life - Numeric
  4. population - Numeric
  5. income - Numeric
  6. region - Character
- Dataset has Population, Life Expectancy, GDP per Capita for Seven regions along with their Countries details for year starting from 1800 to 2015. Follwing are details :
  1. Region : There are seven regions - South Asia, Europe & Central Asia, Middle East & North Africa, Sub-Saharan Africa, America, East Asia & Pacific
  2. Country : Each Region has its Countries as follows. America has 38 Countries. East Asia & Pacific has 31 Countries. Europe & Central Asia has 51 Countries. Middle East & North Africa has 20 Countries. South Asia has 8 Countries. Sub-Saharan Africa has 49 Countries.
  3. Year : There is data from Year 1800 to 2015
  4. Poulation : With minimum value as 1548 and maximum value as 137608943
  5. GDP Per Capita : With minimum value as 142 and maximum value as 182668
  6. Life Expectancy : With mininum value as 1 and maximum value as 84.10
- Data wrangling and Pre-Processing
  1. There are 25817 NA Value in Population column, because population is calculated every ten years. Thus the same population count is applied to every year during the 10 year period.
  2. There are 2341 NA value in income varibale which is only 5.6% of total data thus omitted NA from dataset.
  3. Filtering data for South Asia Region since year 1990
  4. Calculate median of GDP per Capita for each region
  5. Get total population since 1990 for each region
  6. Filtering data for selective year and America, South Asia region
  7. Filtering data for Year 2015
  8. Filtering America region data for Year 1900

Explore Data Analysis and Visualization

Exploratory Plots

In Figure 1, Bar chart visualization showing GDP per Capita for each region starting from year 1800 till 2015. GDP per capita placed at y-axis of chart and each region bar represent by different color over x-axis. Region Color representation can be seen from visualization legends. Visualization shows Europe & Central Asia, America, Middle East & North Africa are top three leading regions with highest GDP per capita. And other regions East Asia & Pacific, South Asia, and Sub-Saharan are arranged on third, fourth, fifth, sixth position respectively.

In Figure 2, stacked bar chart visualization showing population by region for year 1990-2015.Total population placed at chart y-axis where each region stack bar represent by different color over years. Region Color representation can be seen from visualization legends. Visualization shows the total population increase over years for each region. Since 1900 East Asia & Pacific is most populated and Middle East & North Africa is least populated in comparision of other regions.

In Figure 3, Box plot visualization showing Life Expectancy by region. Life Expectancy placed at chart y-axis where each region box plot represent by different color over x-axis. Region Color representation can be seen from visualization legends. Visualization shows two regions - America, Europe & Central Asia’s median Life Expectancy is very near to overall average Life Expectancy.Also each region box plots in these examples show different distributions of Life Expectancy. Here, pattern says South Asia and Sub-Saharan Asia plots are two least distributed regions as compare to other regions.

In Figure 4, Line chart visualization showing South Asian population by region for year 1990-2015. Total population placed at chart y-axis where each country line represent by different color over years. Region Color representation can be seen from visualization legends. Visualization shows the total population increase over years for each countries. India population exponentially increased after 1950 year and hence India became most populated South Asian Country. SriLanka and Pakistan’s population also increased after 1950 but near to year 2000 there is little drop. Afghanistan is the least populated among all South Asian countries.

In Figure 5, scatter plot shows for every 10 years, how Life Expectancy changes for GDP per Capita. Life Expectancy placed at chart y-axis where two region America & South Asia dots represent by different color over GDP per Capita placed at x-axis. Region Color representation can be seen from visualization legends. Since 1980 it can be seen America GDP is start increasing with Life Expectancy. South Asia GDP is vey less correlated with Life Expectancy

In Figure 6, bubble plot shows how each region’s Life Expectancy changes for GDP per Capita for year 2015. Life Expectancy placed at chart y-axis with GDP per Capita at x-axis where each region displayed by different color in form of bubble and bubble size represent population. Region Color representation can be seen from visualization legends.Each region GDP per Capita is correlated with its Life Expectancy. Another observation says,a big group of people are there from South Asia with Life expectancy around 64-70 and GDP per Capita near to $10,000. Similarly, for East Asia & Pacific many people has Life expectancy near to 74-80 and GDP per Capita greater than $10,000

Cluster Aanalysis

Here Average Silhouette method has been used to get number of clusters. Optimal number of clusters (k) is one with highest average silhouette values in plot thus number of cluster are : 4.

Now as we got number of clusters as 4, let’s visualize the result using fviz_cluster function. Here as per clusters count, four groups has been made where each group has different color. Group color representation can be seen in legends.

PCA Analysis

Categorical columns has been removed and data scailing perfored for PCA analysis.Screen plot has been used to plot relative importance of prinicipal components.

It can be seen Comp1 (first prinicpal componenet variance) has largest bar. Let’s extract results using get_pca_var() function and visualize it. Life is most important in explaining the variability in data set. Other variables Income (present in Dim.1) and Population (present in Dim.2) are also contributing in explaining the variability in data set.

Summary

In this analysis, six regions along with their countries have been compared in terms of Life Expectancy, population and GDP per capita over the year 1800 to 1910.The analysis says Europe & Central Asia leading with highest GDP per Capita and Life Expectancy as compare to other regions.East Asia & Pacific is the most populated region and stands at second position for GDP per Capita and Life Expectancy.On drilling down data to compare South Asia and America regions over Life Expectancy and GDP per Capita growth, it can be seen.Since 1980 America GDP started increasing with Life Expectancy however South Asia GDP is almost negligible varied with Life Expectancy.In South Asia, most populated countries are India & Sri Lanka.

References

Phillips, N. D. (2016). Yarrr! The pirateâs guide to R. Available at https://bookdown.org/ndphillips/YaRrr/
Peng, R. D. & Matsui, E. (2018). The Art of Data Science: A Guide for Anyone Who Works with Data. Skybrude Consulting, LLC. Available at http://leanpub.com/artofdatascience
Peng, R. D. (2016). Exploratory Data Analysis with R.Available at https://bookdown.org/rdpeng/exdata/
Grolemund, G., & Wichham, H. (2018). R for Data Science. Available at https://r4ds.had.co.nz/index.html