@somoykhan/binary-decision-tree v1.0.1-alpha1

Binary Decision Tree

A small utility package that takes a Binary Decision Tree as a JSON parameter and executes binary decision tree to return the result. The primary goal is to have a declarative way of writing binary decision trees rather than ugly nested if-else conditionals.

Installation

    npm i @somoykhan/binary-decision-tree

Usage

    const dtCore = require('@somoykhan/binary-decision-tree') 

    const tree = {
    label: 'is somoy > tahmid ?',
    data: { n: 1 },
    evalF: functionForExecution,
    children: [
      {
        label: 'is somoy.age > 21 ?',
        data: { age : 22 },
        evalF: functionForExecution2,
        children: [
          { ... }, { ... } ]
      }
    }

    const executeBDT = dtCore(tree)

    console.log(executeBDT)

Implementation

The current implementation takes a following JSON structure where the children array will take 2 same type of object and so on. The first object of children array will be traversed if the root node evalF return true otherwise the second object of children array will be traversed :

{
    label: 'is somoy > tahmid ?',
    data: { n: 1 },
    evalF: functionForExecution,
    children: [
      {
        label: 'is somoy.age > 21 ?',
        data: { age : 22 },
        evalF: functionForExecution2,
        children: [
          { ... }, { ... } 
        ]

The execution of this object expects:

• label: String
• data: Object
• evalF: Function
• children: Array

The logic behind the execution is simple, is evalF function evaluates true, it will traverse the first object of children array, otherwise it will traverse the second object. And it will keep traversing until the children property is found null, which are the leaf nodes.

Pros

• Configurable
• Simple

Cons

• No Fuzzy Logic, only works in binary decision trees