Thesis Ideas

[Tuomas Peltola]

Hello!

I'm an M.Sc. student in Finance and Data Analytics. Its time for me to start writing my thesis and I've thought a lot about possible topics. P2P lending is starting to look like a likely topic I'm going to concentrate in. I did some research about data availability and happened to run into Kiva. After exploring more about Kiva, I think it would very interesting to focus my research into it. There seems to be a lot of data available, at least in kivatools.com and maybe somewhere else as well. The problem I'm having is that the research problem is much harder to from. Originally, my idea was to predict with classification models whether a P2P borrower would default or not. However, after reading the topics here, it seems like Kiva doesn't provide the information about defaults for everybody. One idea I thought about was predicting whether a borrower will get funded and focus on the determinants of that. I would like to ask whether you guys had any potential suggestion for my Thesis topic? Machine learning and predicting are something I'm especially interested in. I would appreciate any ideas, thanks!

Best regards,

Tuomas 

[Vijay]

Hello,

There is a lot of interest among Kiva lenders in predicting two items - when/if a loan will get

funded and if a borrower would default. I'm sure if you ask in the forums, you can confirm this.

That said, Kiva's data dumps aren't enough for this kind of analysis. You'd  have to get in touch

with Kiva and ask them for more data. I do not know how this works (I assume MBAs and lawyers

are involved more than engineers), but Kiva does provide the data for research.

Some random ideas:

1. you could maybe analyze what parameters help in loans getting funded the fastest?
   given two similar sized loans, why does one get funded in a day and the other takes 25 days?
   there are obvious answers (photograph quality, description quality, gender, location etc), but
   it might be fun to see analysis of it
2. External factors - currency issues, method of payment issues etc - how do they impact repayments?
3. Related to #2, can blockchain/crypto help? For example, with Nano, we can transfer funds worldwide
   in an instant. I have no idea about the viability for Kiva's use case, but there are businesses that
   take payments in crypto and use blockchain for everything from inventory analysis to land ownership.
   
4. Partners - Kiva rates partners, but it would be interesting to also have a third party look into partner data
   
5. community participation - you could get basic teams data. Maybe you could ask Kiva for
   team messages and look into those?

You could check with the captains team, there are amazing people there with deep knowledge and passion.

Once you have chosen the topic, please post to this group :)

Vijay.

--
You received this message because you are subscribed to the Google Groups "build-kiva" group.
To unsubscribe from this group and stop receiving emails from it, send an email to build-kiva+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/build-kiva/424632b7-03e3-4efd-aace-da7069ad57fd%40googlegroups.com.