{"id":3539,"date":"2022-05-27T05:43:30","date_gmt":"2022-05-27T05:43:30","guid":{"rendered":"https:\/\/beeeye.com\/?p=3539"},"modified":"2022-05-27T05:47:12","modified_gmt":"2022-05-27T05:47:12","slug":"decision-making-based-on-multiple-models","status":"publish","type":"post","link":"http:\/\/beeeye.com\/decision-making-based-on-multiple-models\/","title":{"rendered":"Decision making based on multiple models"},"content":{"rendered":"

Decision making based on multiple models<\/h1>\n

Is it possible to design a decision engine based on multiple models? This is the question one of my customers asked me recently. In the world of traditional tools and methods such a requirement calls for some special considerations and new design for the infrastructure you are using. The answer for us using the EyeOnRisk<\/a> platform is of course YES<\/strong>. I will explain exactly why and how in this short post.<\/p>\n

When do we need multiple models?<\/h2>\n

It’s often the case that you wish to segment the population you are trying to score to smaller segments which create more accurate models. A simple example may be to segment population by age: for consumers younger than 25 I will have a pre-trained model and another one for the ones who are older. The models may reveal that the older population holds less risk thus can be granted with a more appealing offer in terms of interest and total amount for a loan.<\/p>\n

To handle such a scenario in your modeling solution would usually require to develop to distinct models, each one developed with the same laborious process either manually or with the aid of a modelling tool<\/a>. Eventually you will need to implement business logic which will act on each record differently according to the AGE of the consumer. Such business logic can be developed easily using the EyeOnRisk platform under the decision engine<\/a> module in the form of a conditional rule set.<\/p>\n

Conditional Rule Set<\/h2>\n

One of the building blocks of a rule set allows to add a conditional block to the logic. This allows to steer the decision based on the value of one or more variables in the input record. In our example above we wish to add a condition based on the value of AGE (which we assume is passed as an input to the decision process). This provides us with the branches needed to differentiate between two models. Assuming we already developed the models and they are ready in the system, we can now activate the appropriate model to calculate the resulting score:<\/p>\n

\"Model<\/a><\/p>\n

Using multiple attached flows<\/h2>\n

Models in the EyeOnRisk platform are implemented using a flow<\/a>. Later this flows can be used in many areas of the system to achieve accurate predictions of your business. In the example above we’ve created two flows which represents the two segments<\/a> of the population we wish to assess (old people vs young people). When we define the rule set, we declare the used flows as “attached flows”. This means that for every activation of the rule set, these flows will be calculated and provide a score based on all the relevant inputs from the record.<\/p>\n

For each attached flow we provide a different target score variable so we can later distinguish between them. These scores are actually used on the right side in the above image, where you can see the conditional statement in the rule set (“if” statement) which takes the proper score according to the AGE variable. In each branch of the if statement the appropriate score is assigned to the final variable holding the score (simply called “score”).<\/p>\n

Conclusion<\/h2>\n

Implementing conditional logic and using multiple models is easy provided that you are working with the right tools. To learn more about the different ways the EyeOnRisk platform can boost your business and time to market – contact us<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"

Decision making based on multiple models Is it possible to design a decision engine based on multiple models? This is the question one of my customers asked me recently. In the world of traditional tools and methods such a requirement calls for some special considerations and new design for the infrastructure you are using. The answer for us using the EyeOnRisk platform is of course YES. I will explain exactly why and how in this short post. When do we need multiple models? It’s often the case that you wish to segment the population you are trying to score to smaller segments which create more accurate models. A simple example may be to segment population by age: for consumers younger than 25 I will have a pre-trained model and another one for the ones who are older. The models may reveal that the older population holds less risk thus can be granted with a more appealing offer in terms of interest and total amount for a loan. To handle such […]<\/p>\n","protected":false},"author":4,"featured_media":1640,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"image","meta":{"_mi_skip_tracking":false,"footnotes":""},"categories":[12,22],"tags":[],"_links":{"self":[{"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/posts\/3539"}],"collection":[{"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/comments?post=3539"}],"version-history":[{"count":3,"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/posts\/3539\/revisions"}],"predecessor-version":[{"id":3543,"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/posts\/3539\/revisions\/3543"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/media\/1640"}],"wp:attachment":[{"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/media?parent=3539"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/categories?post=3539"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/beeeye.com\/wp-json\/wp\/v2\/tags?post=3539"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}