Sample Models

The standard Rule Learner’s installation includes working models that can be used to learn the ML+BR techniques and to create your own machine learning projects.  The models vary from very simple like “Robots” and “Votes” to more complex like “Diabetes” and “Soybeans”. The data for many of these models is copied from the open-sourced WEKA and refactored to Excel. Here is the list of selected ML models:
* Robots    * Lenses    * Credits
* Diabetes   * Labor   * Soybeans    * Votes

ML Model “Robots”

This problem is described at the DMCommunity Challenge Jan-2019. Cyber police has images of Unfriendly and Friendly Robots:

Rule Learner should discover the rules that classify new robots as friendly or unfriendly. Here is the glossary:

Here are the generated rules:

ML Model “Lenses”

Let’s say you already have many cases that describe which contact lens  were prescribed to different people. Rule Learner should use this information to discover human-understandable and machine-executable rules that will allow you to prescribe correct lens to other people. The complete solution is described here.

Here is a fragment of the training instances:

Here is the glossary:

Here are the generated rules:

ML Model “Credits”

This model uses hypothetical credit data set which contains sample of 1000 debtors classified as “good“ or “bad“.  It’s Glossary looks as follows:

The Excel table “instances” contains 1,000 training instances. Rule Learner generates rules that specify the learning attribute “Classified As”:

The detailed description of this model can be found here.

ML Model “Diabetes”

This model is supposed to figure out if someone is likely to have diabetes just by taking a few of these measurements:

Here is the glossary:

There are 768 training instances. Here are the discovered rules:

ML Model “Labor”

This model based on different attributes included in the labor negotiation contracts is supposed to decide if a contract is good or bad. Here is the glossary:

Here are the generated rules:

ML Model “Soybeans”

This model include a large dataset of training instances that describe properties of a crop of soybeans. There are 35 nominal attributes and 683 training instances. Rule Learner is supposed to predict which of the 19 diseases the crop may suffer.

Here is the glossary:

The ML algorithm C4.5 generated 28 rules which correctly classified 656 out of 683 data instances (click on the table to enlarge it):

The ML algorithm RIPPER generated 76 rules which correctly classified 663 out of 683 data instances (click on the table to enlarge it):

ML Model “Votes”

This model based on the voting history of different congressmen should recognize their party affiliation.

Here is the glossary:

Here are the generated rules: