Process mining is an invaluable methodology that extracts insights from event log data to monitor and enhance processes. However, a common point of confusion is that users often believe process mining and diagnostics are one and the same.
We want to clarify that process diagnostics is a specific method within process mining which reveals insights about the problems and deviations within a process.1 As a result, it helps business leaders and experts develop appropriate solutions for the issues in their day-to-day processes.
But first, let‘s quickly compare process diagnostics to overall process mining:
While process mining extracts insights from event logs to understand, monitor, and improve end-to-end processes, process diagnostics focuses specifically on identifying problems, deviations, and opportunities for optimization. It is an analytical technique within process mining.
Process diagnostics has three main applications:
- Problem classification
- Root-cause analysis
- Process trend analysis
However, it‘s important to note that process mining users do not directly use process diagnostics or these applications, but rather the process mining software that enables these capabilities. The software provides the algorithms, visualizations, and analytics to uncover the insights.
Let‘s explore each of these use cases and applications in more detail:
1. Problem Classification
Problem classification refers to predicting categorical classes for inputs or data varieties to generate a model, such as a logistic regression2 or a decision tree3. The classification model categorizes inputs and concludes a classification result by predicting outcomes from them.
Machine learning algorithms facilitate the automatic classification of problems by running these models. For example, 78% of businesses leverage machine learning for security analytics and identifying threats.4
In process mining, classification models help classify issues and deviations detected during conformance checks. Conformance checks compare the observed event log behavior to the modeled expected behavior to detect deviations.
For example, a compliance company’s responsibility is to check if the process cases that it reviews are fraudulent or authorized. In this use case, compliance teams can use process mining‘s classification capabilities to seamlessly separate the legitimate cases from the illegitimate ones.
Special algorithms for classification like decision tree, random forest, logistic regression, and neural networks can be used.
According to one survey, 46% of process mining users leverage decision tree algorithms and 41% use neural networks for classifications related to process diagnostics.5
2. Root-Cause Analysis
Root cause analysis helps businesses find the underlying cause behind problems and deviations by listing the potential reasons and data-driven correlations between them.
Some process mining software solutions offer automated root cause analysis capabilities. This enables users to automatically obtain root causes for each deviation, delayed activity, or error discovered during conformance checks and performance analysis.
Automated root-cause analysis applies machine learning algorithms to diagnose factors that might impact process tasks and operations. For example, it can uncover customer dissatisfaction by analyzing event log data containing information about customer preferences and journeys.
According to one survey, 66% of process mining practitioners utilize root cause analysis, with 15% using automated machine learning-based techniques.
3. Process Trend Analysis
Trend analysis enables identifying patterns in data and understanding changes over time. For example, sales metrics and financial KPIs commonly use trend analysis to detect performance shifts.
When applied to process data, trend analysis reveals principles and changes in process variables. It diagnoses and controls processes through a data-driven approach.
In essence, process trend analysis enables data-driven decision making by improving process supervision compared to traditional visual and human-driven methods.
It pinpoints positive and negative patterns, so analysts can define high-performing processes as standard models while improving underperforming operations.
Process mining attributes like performance analysis and conformance checks enable effective trend analysis. For example, a global bank uses process mining to analyze trends in their account opening process. This helped them flag an unexpected loop that was generating customer complaints and rework cases.6
Applications of Process Diagnostics
Process diagnostics enables optimization across many industries and use cases, including:
-
Healthcare: A hospital in Spain used process mining diagnostics to analyze their cancer patient journey. It reduced diagnosis delays by identifying trend deviations and bottlenecks.
-
Supply chain & logistics: Trend analysis monitors complex distribution processes to detect inefficiencies for proactive optimization.
-
Finance: Problem classification accurately flags fraudulent transactions. Root cause analysis identifies what factors enable fraud.
-
Insurance: Algorithms classify claims as high or low risk to streamline processing and reduce leakages.
-
Manufacturing: Machine learning pinpoints reasons for production downtime, scrap, or recalls for waste reduction.
The Bottom Line
To summarize, the top 3 process diagnostic applications are:
- Problem classification – Categorizing issues using models like logistic regression and decision trees
- Root cause analysis – Finding underlying reasons for problems and deviations
- Trend analysis – Identifying patterns and changes in processes over time
Process mining software provides the algorithms, visualizations, and analytics to enable organizations to leverage these diagnostic capabilities for data-driven insights.
By thoroughly understanding processes, exceptions, and trends, process diagnostics empowers businesses to optimize operations, increase efficiency, and exceed outcomes.