An exploratory analysis on Kiva loan data found here on Kaggle.com.
- Subhashini Chodavarapu
- Min Kim
- Tania Mukherjee
- Sadhana Sankar
-
Overall picture of the statistics of kiva loans all over the world(eg: loan_amount, gender split, activity, sectors, etc)
-
Have a base world map which can be zoomed on each continent and be able to click on each country which would pop up the existing statistics on Kiva data Countrywise (eg: loan_amount, most_active_sector )
-
Kiva growth over the years (Number of loans/Total loan amount over the years, stacked by sector)
-
Gender disparity by regions (Map) & sectors, overall growth through time(Pie/bar/line charts)
-
Overall number of loans, sum of loans, Loans by sector (different layers in map)
-
Timeline (animated)… Loans by time , color coded by sector, size by loan amount.
-
Have a base world map which can be zoomed on each continent and be able to click on each country which would pop up the existing statistics on Kiva data Countrywise (eg: loan_amount, most_active_sector )
-
Show correlation between gender and kiva loans and come up with other correlations (we will work on this using tableau and come up with insights)
-
Where could the money lended by lenders be efficiently utilized ? (Measure by growth/poverty)
-
Deeper look at Philippines, Kenya & Peru (Top 3 by number of loans) By sector, activity, repayment. Scrape local news around the time of much increase in loans (Is it Kiva popularity or is it the need?) .
Deeper look at Last three, budding regions. Scope for more funding? (How much do they need really?)
Mid three? How can we improve ? Do they need? (Verify with poverty index )
-
Lending to teams vs Lending to individuals, in what all sectors does it make a difference?
-
Predict where could the money lended by lenders be efficiently utilized (to be decided)
-
Will a loan money returned, be refunded into other loans?? (How to maximize refunding? Where they refund ? same sector?)
-
How many borrowers might come forward to fulfill a new loan?
-
Group the loans or borrowers into clusters by Unsupervised learning techniques and identify top n features (K-Means)