🛑Decision automation should N🚫T be either #MachineLearning or #BusinessRules❗

💢Often, teams developing solutions to automate business decisions think about using technology such as machine learning #ml #data over business #rules. These two options are not alternatives to each other. They should be used together as they solve different types of problems.

✨Firstly, decision automation is about business decisions. So, you should not start with either machine learning or business rules. It is about business decisions, so let's start with framing a business decision. Model the holistic view of a business decision to create a hierarchical Decision Graph with connected nodes of decision units and data inputs.

You can now reap the benefits of this #decisioncentric approach as the big problem is decomposed into bite-size pieces. Based on the nature of every decision unit, choose the right technique, i.e., machine learning AND business rules.

1️⃣Machine Learning: For decision units that utilize machine learning, the whole process of building the data pipeline, cleaning, model training, model optimization, and so on is easier. Why? because it is a small model focused on a specific decision unit and has limited scope. Therefore, finding data is easier, training is faster, and the trained model will be more accurate.

2️⃣Business Rules: In business rule-driven units, all related rules are grouped and encapsulated in reusable modules. Therefore, rules management is straightforward without the need for redundant and extra grouping, labeling, and tagging overhead. Potential future changes are all isolated to specific decision units. Therefore, maintenance is quick, has lower risks than the traditional approach, and has lower costs. Additionally, understanding the changes and rules is easier.

🚀From the overall solution perspective, you used both techniques in conjunction. The decision graph, which represents the holistic business decision, is a self-sufficient model. It can be deployed and managed through the entire lifecycle of the change management process in the organization. Integrating it into the organization's processes and systems is easier because it has clear boundaries.

#decisionintelligence

Posted here.

Published April 30th, 2024 at 07:30 am