An important result from running a Process Mining analysis is Root Cause Analysis to discover the where and why inefficiencies occur by finding hidden connections and influencing factors in process data.
In PO Academy installment #14, Dan Grady provides a session on Root Cause Analysis, and in this post, we’ll review how the (OR) Odds Ratio (this is x times more likely) is calculated. Below is a detailed view of a Root Cause Analysis result from the Process Optimization Academy session. The value “2.21 times more likely than any other assignment group” is the Odds Ratio.
The Odds Ratio (Wikipedia) is a statistic that quantifies the association or correlation between 2 events, A and B, and is defined as the ratio of the odds of A in the presence of B and the odds of A in the absence of B. Ok, enough with the stats speak, what does that mean in terms of Process Mining?
Using the example above, the Odds Ratio helps us to quantify the relationship between the IT Support Service Desk Level 1 Assignment Group and any other assignment group considering the condition of 12 hours or more between Work in Progress and Resolved. It’s easier to see when we build a contingency table using the 4 incident counts provided by the Root Cause Analysis output;
|
Condition |
||||
|
> 12 hours |
< 12 hours |
Total |
||
|
Assignment Group |
Level 1 |
1978 |
3767 |
|
|
Any Other |
||||
|
Total |
8400 |
22500 |
||
Now we are able to fill in the 5 missing values in the contingency table;
|
Condition |
||||
|
> 12 hours |
< 12 hours |
Total |
||
|
Assignment Group |
Level 1 |
1978 |
1789 |
3767 |
|
Any Other |
6422 |
12311 |
18733 |
|
|
Total |
8400 |
14100 |
22500 |
|
Next, we can calculate the odds of the incident taking >12 hours when it’s Level 1 Assignment Group; 1978 / 1789 = 1.1056. From our definition above, this is the odds of A (an incident taking >12 hours) in the presence of B (Level 1 Assignment Group).
And the odds of the incident taking > 12 hours when it’s any other Assignment Group = 6422 / 12311 = 0.5216. This is the odds of A (an incident taking >12 hours) in the absence of B (any other Assignment Group).
The Odds Ratio is now simply the ratio of these two odds values, 1.1056 / 0.5216 = 2.1, and since the OR is greater than 1, these two events are associated with each other. Meaning compared to any other assignment group, IT Support Service Desk Level 1 is 2.1 times more likely to take more than 12 hours between Work in Progress and Resolve.
The key takeaway is this – we need to be able to quantify the strength of relationships found when analyzing process data to focus our improvement efforts where the greatest potential value is. Root Cause Analysis and the Odds Ratio are two powerful results Process Mining calculates for us to help prioritize improvement initiatives by showing us evidence of associations that are not purely by chance alone.
1 Comment
