Dependency matrix issue

wo7777 · September 2023

I've been playing around with various tools to try and understand / generate dependency matrices.

This is my very small data set :

patient handling handling_id registration_type time employee

test1 C1 NULL9 complete 2023-01-31 13:01:28 NULL

test1 C2 NULL12 complete 2023-01-31 12:41:42 NULL

test1 A1 NULL13 complete 2023-01-31 10:01:46 NULL

test1 I1 NULL14 complete 2023-01-31 10:00:33 NULL

test2 M1 NULL16 complete 2023-02-02 21:58:00 NULL

test2 C2 NULL18 complete 2023-01-31 12:57:44 NULL

test2 A1 NULL20 complete 2023-01-31 10:15:00 NULL

test2 I1 NULL21 complete 2023-01-31 10:14:17 NULL

test3 C2 NULL27 complete 2023-01-31 14:23:47 NULL

test3 A1 NULL29 complete 2023-01-31 10:18:00 NULL

test3 I1 NULL30 complete 2023-01-31 10:17:21 NULL

test4 A1 NULL38 complete 2023-01-31 12:49:00 NULL

test4 C2 NULL39 complete 2023-01-31 12:34:08 NULL

test4 A1 NULL40 complete 2023-01-31 12:34:00 NULL

test4 C1 NULL41 complete 2023-01-31 11:04:33 NULL

test4 A1 NULL42 complete 2023-01-31 10:46:00 NULL

test4 I1 NULL43 complete 2023-01-31 10:45:25 NULL

When I run a dependency matrix I get this :

Edges

# A tibble: 9 x 7

antecedent consequent dep from_id to_id n label

I1 A1 0.8 5 1 4 4

C2 C1 0.5 3 2 1 1

A1 C2 0.5 1 3 5 5

A1 End 0.5 1 4 1 1

C1 End 0.5 2 4 2 2

C2 End 0.5 3 4 2 2

M1 End 0.5 6 4 1 1

Start I1 0.8 7 5 4 4

C2 M1 0.5 3 6 1 1

I was looking at one of the van der Aalst videos on dependency graphs, and I would have thought that if you apply the antecedent/consequent approach to calculate the matrix e.g. as b follows a etc, then the calculation would be dep = a/a+1 but you can see we get values like 0.5 which is basically impossible as I understand it. E.g. A1 --> C2 you have a count of 5 so (5-0) / (5+0+1) = 0.8333, not 0.5

Any help appreciated.

Dependency matrix issue

Howdy, Stranger!

Categories