Book Image

AI and Business Rule Engines for Excel Power Users

By : Paul Browne (GBP), PORCELLI
Book Image

AI and Business Rule Engines for Excel Power Users

By: Paul Browne (GBP), PORCELLI

Overview of this book

Microsoft Excel is widely adopted across diverse industries, but Excel Power Users often encounter limitations such as complex formulas, obscure business knowledge, and errors from using outdated sheets. They need a better enterprise-level solution, and this book introduces Business rules combined with the power of AI to tackle the limitations of Excel. This guide will give you a roadmap to link KIE (an industry-standard open-source application) to Microsoft’s business process automation tools, such as Power Automate, Power Query, Office Script, Forms, VBA, Script Lab, and GitHub. You’ll dive into the graphical Decision Modeling standard including decision tables, FEEL expressions, and advanced business rule editing and testing. By the end of the book, you’ll be able to share your business knowledge as graphical models, deploy and execute these models in the cloud (with Azure and OpenShift), link them back to Excel, and then execute them as an end-to-end solution removing human intervention. You’ll be equipped to solve your Excel queries and start using the next generation of Microsoft Office tools.
Table of Contents (22 chapters)
Free Chapter
1
Part 1:The Problem with Excel, and Why Rule-Based AI Can Be the Solution
5
Part 2: Writing Business Rules and Decision Models – with Real-Life Examples
9
Part 3: Extending Excel, Decision Models, and Business Process Automation into a Complete Enterprise Solution
13
Part 4: Next Steps in AI, Machine Learning, and Rule Engines
Appendix A - Introduction to Visual Basic for Applications

Another machine learning method – decision trees

In Chapter 11, we trained a machine learning model using Naïve Bayes. At the time, we mentioned that we only needed to change one line to swap in other machine learning methods. If you take a look at the 12_tree.ipynb notebook on the book’s GitHub page, you’ll see we’ve done exactly that. The line we create our classifier in the notebook now reads as follows:

model = PMMLPipeline([( "classifier", DecisionTreeClassifier(),)])

Running the notebook in Azure or another notebook hosting option is the same as in the previous chapter, so we won’t repeat ourselves. More importantly, when you run the notebook, you get a prediction graph, as shown in Figure 12.1:

Figure 12.1 – Predictions from the decision tree model

Compare this result to Figure 11.16 in the previous chapter. While the predictions are broadly similar, they are not as fine grained. This...