The Rules screen displays all of the predictive rules for the outcomes in the category that we want to predict. To easily understand the rules, the engine divides them into three different sections based on how often the rules occur within the original data. For the screen shots below, we will be using the diabetes data set, available here:
diabetes_input_data_file.csv.zip
Let's start with the area on the right side of the screen that shows the Targeted Rules.
The Targeted Rules are our rules that have very high probabilities, but have very low frequencies. This means that these items don't come together very often- but when they do, they are highly predictive.
Clicking in the middle of the screen will bring up a list of Common Rules.
The Common Rules occur somewhat frequently, and when they occur they are fairly predictive. These items come together more often than the Targeted Rules, but when they do they may not lead to a targeted outcome as often.
Clicking on the left side of the screen will display the Obvious Rules.
The Obvious Rules occur very often throughout the data set, but may not be as predictive as our other two rules, and will typically have lower Conditional Probabilities.
When making real-time predictions, our engine uses combinations of many different rules to make predictions, so each rule that is seen in the Rules screen is impactful in some way. For more information on our Real-Time predictions engine, click here.
