rawles/kitti

Klau's Intricate Transactions Totting Index

kitti's Issues

Run t-SNE on transaction graph to better show relationships

Current circle distribution is now hard to read.

Feature: who should pay?

It's a slightly ill-defined problem. Here are some different ways of looking at it:

1. Choose the person who is overall most in debt to kitti. This might be appropriate if the whole group mixes randomly.
2. Choose the person who is most in debt with the people present at the event. A better solution if some of the people present have different levels of interaction with the group. Also, as the graph grows this becomes a more natural solution.
3. Choose the person who most improves the kitti 'health' (expected number of transactions for network to clear based on previous transactions). This encourages people on the edge of the graph to clear their debt sooner. This is harder to reason about and may not correlate with social expectations.

Feature: Click on a person to see their transactions only

User story:

Why am I in debt? I don't remember what happened. Please show me the transactions that involve me.