Process Mining and Simulation
Ha Eun Kim| Mar 7 2022| 8 min read
Process mining, when combined with simulation, offers a powerful method for understanding, analyzing, and optimizing complex processes by visualizing current states, tracking key performance indicators, and evaluating various improvement scenarios to derive the best solutions with limited resources.
Are there any ‘processes’ around you?
According to the dictionary, ‘process’ is defined as the path or sequence in which something is handled. You may have heard or experienced terms like ‘product production process,’ ‘logistics process,’ or ‘business process.’ Nowadays, most processes leave traces in systems such as ERP, CRM, MES, making it easy to track their progress.
‘Process mining’ is an analysis technique that uses event logs to accurately understand the flow and status of processes. Furthermore, when combined with ‘simulation,’ which allows you to validate improvements in virtual reality, you can effectively identify issues and choose the best solutions with limited resources.
In this session, we will briefly explore how process mining can help understand the current state of processes and how incorporating ‘simulation’ can lead to the discovery of process improvement ideas.”
Understanding the Current Process
First, you’ll need data such as event logs. The data should contain information about the actors in the process, what actions were taken, and when they occurred.
Shall we take a look at the process of a patient visiting a hospital? Let’s consider the process of a patient named ‘Logan’ who arrived at the hospital in the morning.
‘Logan’ checked in at the reception, followed the instructions to have their blood pressure measured, waited for their turn, and then received a medical examination. Afterward, they went to the payment counter to settle their medical fees.
The data for ‘Logan’ is stored in the hospital system, allowing us to gather the following information:”
The data can be visually represented as follows:
The time taken for registration, blood pressure measurement, consultation, and payment is 10 minutes each, and the waiting times are 10 minutes, 35 minutes, and 35 minutes, respectively. It seems that the hospital had a lot of patients today, and the consultation and payment took a long time.
This time, I want to analyze the entire patient visit process, not just one person’s data. Just like the data for the patient ‘Logan,’ you can gather data for other patients as well.
Just as you did before, you can create a map for each patient’s visit process.
<Logan’s Process Map>
<Nicole’s Process Map>
<Alex’s Process Map>
Furthermore, you can go a step further and integrate the data from multiple patients into a frequency-based process map.
It’s not just about visualizing the process; you can also calculate and track important Key Performance Indicators (KPIs) that require management. In this example, key metrics like the average time to complete the visit process, average time for registration, and waiting times for each step appear to be crucial.
Here is an analysis dashboard screen constructed with ProDiscovery from Puzzle Data. Using various analysis puzzles, you can perform AS-IS analysis to identify areas that require management.
Generating Process Improvement Solutions
So far, we have examined the current state of the process. Now, we need to derive the most effective solution.
What would happen if a solution was derived without sufficient consideration and applied in reality?
For example, let’s consider hiring new staff to reduce patient waiting times at the payment counter. If a significant amount of time and cost is spent on recruiting and training new staff, and the problem is not resolved, the hospital may need to invest additional time and resources to address the issue.
And when it’s challenging to compare multiple improvement options, determining the best one becomes even more difficult.
In such situations, one of the most effective methods is to utilize “virtual reality.”
Have you heard the expression “run a simulation” before? To be more specific, it involves slightly altering the values of interest in the process to create new data (logs).
By generating simulation logs, you can observe changes in the total process time, task durations for specific steps, waiting times, and idle times of workers. This allows you to understand the impact of proposed improvements on the process. Additionally, from a process mining perspective, you can compare the simulation logs with existing event logs or perform a more detailed analysis of the simulation logs themselves.
|Generate a new event log by changing the person in charge for the “Payment” task|
Let’s examine an example:
There are two potential solutions to reduce patient waiting times at the payment counter:
- Hire a New Employee: You can hire a new employee to assist with the payment process.
- Adjust Work Hours: Instead of having one employee in the morning and three in the afternoon for payment, you decide to change it to two employees in the morning and two in the afternoon.
For each approach, you modify the workforce in the process and generate simulation logs. Next, you compare these simulation logs with the existing logs to assess the differences in key performance indicators (KPIs). You evaluate how much the patient visit process has improved, including the average completion time, payment task duration, waiting times, and overall process costs compared to the baseline. Finally, you select the more effective improvement solution and implement it in the real world.
Indeed, integrating simulation into process mining allows you to not only analyze processes but also compare multiple solutions with limited resources to derive the best improvement plan. The use of simulation helps in evaluating “what-if” scenarios effectively.
The addition of simulation capabilities in ProDiscovery 3.0 by Puzzle Data will be a valuable tool for users looking to perform effective what-if analyses in the context of process mining. It can aid in making data-driven decisions and optimizing processes for efficiency and cost-effectiveness.
In fact, besides what I mentioned today, ProDiscovery 3.0 simulation offers many more useful and exciting features. In our next session, I’ll share another story related to process mining and simulation. Please stay tuned!