PPM with Minitab & Crystal Ball vs. AI for Process Improvement ~ Anurag Fuloria

When it comes to process improvement, two main approaches emerge: Process Performance Modelling (PPM) and Artificial Intelligence (AI). PPM, utilizing tools like Minitab and Crystal Ball, relies on statistical modelling and historical data to understand relationships between variables and predict performance. This offers transparency and interpretability, making it ideal for well-defined processes with good data. AI, on the other hand, leverages machine learning algorithms to learn from diverse data sources, including real-time and unstructured data. While AI excels at automation, optimization, and complex pattern recognition, its "black box" nature can make decision-making less transparent. Choosing the right approach depends on your specific needs. If interpretability and clear improvement strategies are priorities, PPM shines. If automation, complex data analysis, and intelligent recommendations are your focus, AI takes the lead. In some cases, a hybrid approach combining both PPM and AI can be the most effective strategy for optimizing your processes.

For those familiar with the good old CMMI days, creating Process Performance Models (PPMs) using Minitab and Crystal Ball might feel like revisiting an old friend. But for our newer colleagues, here's a simplified breakdown: Imagine you want to improve your incident management process. PPMs, like building a map, help you understand how different factors – like incident priority or agent availability – affect key metrics like resolution time. We use Minitab to analyze historical data on these factors, and Crystal Ball lets us simulate different scenarios (e.g., higher incident volume) to predict potential impacts. This data-driven approach helps us pinpoint bottlenecks and identify areas for improvement, ultimately leading to a smoother incident resolution experience for everyone.

Below steps should provide a comprehensive overview of creating a Process Performance Model (PPM) for the IT incident management process using Minitab and Crystal Ball.

1. Define and Gather Data:

• Identify key variables: Start by defining the critical factors impacting your incident management process. For example, consider variables like: Incident Priority:?High, Medium, Low First Call Resolution Rate (%) Average Time to Resolution (Hours) Number of Escalations Agent Availability
• Gather historical data: Collect historical data for each chosen variable over a specific period. Ensure the data is accurate and representative of your incident management process.

2. Data Analysis & Model Selection (Minitab):

• Import data:?Open Minitab and import your gathered data, creating a dataset with each variable as a column and each incident as a row.
• Exploratory Data Analysis (EDA):?Perform EDA to understand the distribution of your data. Use tools like histograms, boxplots, and scatterplots to identify trends, outliers, and relationships between variables.
• Choose the right model:?Based on your data analysis, select an appropriate statistical model to represent the relationship between your key variables. Common options for incident management include: Regression models:?Linear regression (for continuous dependent variables) or Logistic regression (for binary dependent variables like first-call resolution). Time Series analysis:?If the data shows trends over time, consider autoregressive integrated moving average (ARIMA) models.
• Model fitting and diagnostics:?Once you choose a model, use Minitab to fit the model to your data. Analyze model diagnostics like residuals plots to ensure the model assumptions are met and the fit is appropriate.

3. Sensitivity Analysis & Scenario Planning (Crystal Ball):

• Import model from Minitab:?Transfer your fitted model from Minitab to Crystal Ball. This allows for simulation and scenario planning.
• Define distributions:?Assign probability distributions to each input variable in your model based on your data analysis. Crystal Ball offers various distributions like Normal, Poisson, or user-defined distributions.
• Sensitivity Analysis:?Use Crystal Ball to perform sensitivity analysis. This allows you to see how changes in one input variable (e.g., agent availability) affect the output variable (e.g., average resolution time). Identify the most influential variables through sensitivity analysis.
• Scenario Planning:?Develop different scenarios by setting specific values or ranges for your input variables. Crystal Ball can then simulate these scenarios and predict the potential impact on your incident management process outcomes. For example, simulate a scenario with increased incident volume and reduced agent availability to assess potential impact on resolution times.

4. Model Validation and Interpretation:

• Validate the model:?Compare the actual data with the model predictions to assess its accuracy. Utilize techniques like residual analysis to identify any deviations.
• Interpret results:?Analyze the results of your sensitivity analysis and scenario planning. Identify how changes in input variables affect KPIs like first-call resolution rate and average resolution time. Use these insights to identify improvement opportunities in your incident management process.

5. Model Improvement and Continuous Monitoring:

• Model refinement:?Based on the validation results, you may need to refine your model by adding additional variables, changing the model type, or addressing data quality issues.
• Continuous monitoring:?Monitor your incident management process KPIs and model performance over time. As new data becomes available, consider updating your model to reflect changing conditions and ensure its ongoing relevance.

By following these steps and continuously refining your model, you can create a valuable tool for understanding, optimizing, and predicting the performance of your IT incident management process. Remember to customize the chosen variables and model type to best reflect your specific process and objectives.

Now, let us look into the below table comparing Process Performance Modelling (PPM) using Minitab and Crystal Ball with AI for process improvement:

Choosing the Right Approach:

The best approach for process improvement depends on your specific needs and data availability.

• PPM is a good choice when you have historical data for key process variables and need to understand the relationships between them. It allows for transparent and predictable improvement strategies.
• AI is a powerful option when you have access to large amounts of data (potentially including unstructured data) and want to automate tasks, optimize decisions, or make intelligent recommendations based on complex patterns in the data. However, AI models might be less interpretable.

Consider a hybrid approach if both scenarios partially apply. Leverage PPM for initial analysis and scenario planning, then use AI for specific tasks or automation within the process based on the insights gained from PPM.

Keep learning and sharing!!!

For more on Next-Gen AI Powered SIAM : https://lnkd.in/dUJ-YhNc