Help international clustering project
- 1. HELP International Project Humanitarian Aid to Underdeveloped Countries By: Pikasha Sharma
- 2. Problem Statement Help “HELP INTERNTIONAL” to categories countries based on socio economic factors such as child mortality, income, GDP, life expectancy etc., to optimally and strategically allocate the recent $ 10 million funding that the NGO raised. The aim is to find out the countries in most need of aid based on various socio economic and health factors that will determine the overall development of the country.
- 3. Insights from Exploring Data • From the graph we can see that GDP and Income are highly positively correlated, higher the income of a country higher is the GDP. • High income and GDP also leads to higher expenditure on health facilities by people. • Hence, we can conclude that income and GDP are good measures to indicate the health of a country.
- 4. Insights from Exploring Data • Countries with high export and import indicates robust domestic and international demand, good economic strength and sustainable trade surplus and deficit. Thus, they have a high GDP which can also be seen from the graph. • Inflation rate is very high in poor countries (countries with low GDP) indicative of an unstable economy.
- 5. Insights from Exploring Data • Countries having low income have very high fertility (The number of children that would be born to each woman if the current age-fertility rates remain the same) this maybe due to lack of education on birth control methods. • Rich countries have low total fertility. • Poor countries have high child mortality and low life expectancy.
- 6. Solution for Categorizing Countries Having figured out how various factors affects the overall economic status of a country we want to group similar countries together. This problem can be handled by using clustering algorithm known as k-Means Clustering. K- Means Clustering is an iterative algorithm which tries to partition the data set into K pre-defined distinct and non-overlapping sub groups wherein each data point belongs to only one group. Each data point is assigned to a cluster (sub-group) such that the sum of the squared distance between the data points and the cluster’s centroid (arithmetic mean of all the data points that belong to that cluster) is at the minimum. We will use the features of our data set to find out the subgroups of countries which are similar to each other. For example countries having similar income, GDP, child mortality rate will be clubbed together.
- 7. How does the k Algorithm works? First we decided on the number of clusters the countries would be divided into suppose that’s k. Pick k random points as cluster centers. Assign each data points to their closest cluster centers based on Euclidean Distance. Update the center of each cluster based on the included observations (Mean of all the observations) Terminate if no observations changed cluster, otherwise go back to step 3. After performing clustering countries were divided into 4 clusters:- - cluster 0: very low income and very high mortality rate. - cluster 1: very high income and low mortality rate. - cluster 2: low income and high mortality rate. - cluster 3: high income and low mortality rate.
- 8. Analysis of clusters formed Cluster 0 has the highest child mortality rate. The lowest GDP and It is safe to assume that these countries will also have the lowest income. Countries in these clusters also spend minimum on health. We can say that our countries of interest lie in cluster 0 , as they have the lowest income, lowest GDP and the highest child mortality which is a measure of economic development.
- 9. Countries in Cluster 0 Countries in need of aid Congo, Dem. Rep. Malawi Madagascar Afghanistan Chad Cote d'Ivoire Pakistan Liberia Guinea Comoros Gambia Tanzania Ghana Yemen Burundi Togo Eritrea Kiribati Senegal Zambia Nigeria Niger Sierra Leone Burkina Faso Benin Lesotho Mauritania Congo, Rep. Central African Republic Rwanda Haiti Timor-Leste Kenya Sudan Angola Mozambique Guinea-Bissau Uganda Mali Cameroon Lao Namibia • Most of the countries in cluster 0 belong to the Sub-Saharan Africa region. • All of these countries have high child mortality, low income, low GDP, low life expectancy and very high inflation.
- 10. Top 5 Countries in dire need of aid • Child mortality for these countries is above 85th percentile. • Income of these 5 countries is below the 3rd percentile mark. • They also have very low life expectancy (less than 18th percentile i.e., 61 years).
- 11. Conclusion Congo, Liberia, Burundi, Niger, Central African Republic require aid most dearly. They all have extremely low income and GDP which is an important measure of economic status of a country. They also have very high child mortality this will be because of lack of basic medical facilities in these countries. These countries also have very high inflation rates with poor import and export.