What kind of insights can we give from point A to point B transactions. A and B are two places, All I can think of is about descriptive statistics like how many transactions from A and what is count of transactions between places and all.
Asked
Active
Viewed 41 times
1
-
do you have timing information when the transaction occur? – nairboon May 14 '19 at 08:13
-
Yes i have date of transaction, can i do something with date time and material moved/vehicle used? – Pardhu May 14 '19 at 09:11
-
how many different A's and B's are there? – nairboon May 14 '19 at 11:40
-
around 50k but 700 unique routes. like if A is Pune and B is Nagpur then the same A and B for different prices and different times 300 times. – Pardhu May 14 '19 at 11:43
1 Answers
1
You could use the tools from Network Science (1) to model (A,B) as edges in a network. Since you have timing information on the transactions you can reconstruct paths on that network.
Using statistical network analysis you can find for example which points A from your dataset (or nodes in the network) are of importance, as in how many routes lead to that point. For this you'd study a centrality measure (degree centrality, closeness centrality, betweenness centrality, etc.) (2)
Or you can find which points are close together in terms of transactions between them, for this you'd look for community detection or clustering. (2)
You can also look at the timing data, to calculate the intra- and inter- transaction time distributions.
(1): Barabási, A. L. (2016). Network science. Cambridge university press.
nairboon
- 123
- 3