ProcessMIX Guide
  • Platform Overview
    • Introduction
    • Platform Purpose
    • Main Concepts
      • Organization Structure
      • Visual Language
      • Project Development Environment (IDE)
      • Deployment and Integration
      • Administration and Troubleshooting
    • First Steps
    • Change Log
      • 5.0.0
      • 5.0.5
      • 5.0.5.1
      • 5.0.5.2
      • 5.0.5.3
      • 5.0.5.4
  • Language Reference Guide
    • Overview
    • Data Structures and Types
      • Core Types (pdk.core)
      • Dictionary (pdk.asset.dictionary)
      • Scorecard (pdk.asset.scorecard)
      • Blockchain (pdk.blockchain)
      • Exceptions (pdk.ex)
      • Input/Output (pdk.io)
      • JSON Web Token (pdk.jwt)
      • Networking (pdk.net)
      • HTTP (pdk.net.http)
      • SOAP/Web Services (pdk.net.soap)
      • XML/DOM (pdk.net.xml)
      • S3 (pdk.s3)
      • SQL (pdk.sql)
      • Postgres (pdk.db.postgre)
      • Util (pdk.util)
      • Default Data Structure Field Values
    • Flows
      • Variables
        • Local Flow Variables
        • Node Variables
      • Flow Node Types
        • Core Nodes
          • In
          • Out
          • Assign
          • If
          • Switch
          • Iterator
          • Function Call
          • Subflow
          • Raise Exception
          • Exception Handler
          • Validator
        • Connector Nodes
          • Database Begin Transaction
          • Database Commit Transaction
          • Database Rollback Transaction
          • Database Query Executor
          • EVM Blockchain Functions
          • EVM Blockchain Smart Contract Functions
          • REST Service Executor
          • SOAP Service Executor
          • S3 Connector
          • SMB Connector
      • Expressions
      • Built-In Functions
        • Expression functions
          • pdk.util.Any
          • pdk.util.Array
          • pdk.util.Blockchain
          • pdk.util.Cast
          • pdk.util.Codec
          • pdk.util.Crypto
          • pdk.util.Date
          • pdk.util.File
          • pdk.util.Json
          • pdk.util.JWT
          • pdk.util.Map
          • pdk.util.Math
          • pdk.util.String
          • pdk.util.Util
        • Node functions
          • pdk.flow.Any
          • pdk.flow.Array
          • pdk.flow.File
          • pdk.flow.JsonArray
          • pdk.flow.JsonObject
          • pdk.flow.HttpRequest
          • pdk.flow.HttpResponse
          • pdk.flow.Logger
          • pdk.flow.Map
          • pdk.flow.Xml
            • pdk.flow.xml.XmlAttr
            • pdk.flow.xml.XmlElement
            • pdk.flow.xml.XmlNode
            • pdk.flow.xml.XmlDocument
            • pdk.flow.xml.XmlDocumentType
            • pdk.flow.xml.XmlDOMConfiguration
            • pdk.flow.xml.XmlDOMImplementation
            • pdk.flow.xml.XmlProcessingInstruction
            • pdk.flow.xml.XmlTypeInfo
            • pdk.flow.xml.XmlCharacterData
            • pdk.flow.xml.XmlText
            • pdk.flow.xml.XmlNamedNodeMap
          • pdk.node.BlockchainEth
          • pdk.node.S3
          • pdk.node.Smb
      • Exceptions
        • Checked Exceptions
        • Runtime Exceptions
    • Assets
      • Dictionary
      • Decision Table
      • Scorecard
      • PMML
    • Connectors
      • DB Connector
        • Prepared Query
        • Dynamic Query
        • Query input parameters
        • Stored Procedure
      • REST Connector
      • SOAP Connector
      • S3 Connector
      • SMB Connector
      • EVM Blockchain Connector
    • Global Variables
    • Appendices
      • Overview of Database Transactions
      • Reserved Words
  • Project Development Environment (IDE)
    • Project Explorer Panel
    • Flow Builder
    • Flow Node Editor
    • Data Structure Builder
    • Connector Builders
      • REST Connector Builder
      • SOAP Connector Builder
      • DB (Database) Connector Builder
      • EVM Blockchain Connectors Builder
      • S3 Connector Builder
    • Asset Builders
      • Dictionary Builder
      • Decision Table Builder
      • Scorecard Builder
      • PMML Asset Builder
    • Global Variables Panel
    • Expression Editor
    • Project Deployment and Execution
    • Cron Expression Generator
    • Test Helper
    • Debugging the Project
    • DB transactions
    • Team Collaboration Tools
      • Version Control
      • Conflict Resolver
    • Selectors
    • Error Panel
    • Local History
  • Home and Administration Guide
    • Organization and Subscription
    • Organization Team
    • Repositories and Projects
    • Environments
    • Deployments
    • Database Provisioning
    • Roles and Permissions
  • Logs and Troubleshooting
    • Request/Call Logs
    • Deployment Logs
    • Application Logs
  • Appendix: Example Project
    • Risk Mitigation Solution
Powered by GitBook
On this page
  1. Language Reference Guide
  2. Assets

Scorecard

PreviousDecision TableNextPMML

Last updated 6 months ago

A scorecard is a statistical or machine learning predictive model used primarily in fields like credit risk, marketing, and healthcare to predict future outcomes based on historical data. It assigns a numerical score to a set of characteristics or inputs, which represent the likelihood of a specific outcome—such as a customer’s likelihood to default on a loan or to respond to a marketing offer.

A scorecard takes a set of predictive parameters, called characteristics. The entire range of possible values for a given characteristics is split into several sub-ranges, called bins. Each of the bins gets assigned a numeric value—a partial score. When a data request is processed by the scorecard, it checks for each parameter which bin it falls into it and extracts the partial score for it. Then it summarizes all partial scores to produce one value—the final score, which can also be further processed, or normalized, to make it fall within a predetermined range of values. With the help of the final score, all requests can be rank-ordered by their likelihood to produce a desired outcome for the business. Typically, the higher the final score, the better.

For example, a scorecard can try to predict the likelihood of repaying a loan. It can pick some of the parameters from the loan application request, such as income, age, number of past bankruptcies or past due payments, etc. as predictive characteristics.

Each bin in a characteristic may be assigned a reason code and message. When a final score is calculated, the scorecard will automatically produce a list of reason codes, rank-ordered by their contribution (partial score) to the final outcome. This way it is easy to demonstrate to clients what influenced the score-driven decision—make it clear, for example, why their loan application was declined.

Typically scorecards are trained automatically using statistical methods on historical (already known) outcomes, but you can also develop them by hand to capture expert-driven decision strategies.

When a scorecard asset is created in ProcessMIX, a backing to hold its request parameters is auto-generated for it as well. As you edit the scorecard parameters, the system automatically updates the supporting data structure's properties to keep it in sync.

The reason codes for a scorecard asset, if they are used, come from a asset. It has to be configured to have a "Reason code" type, so that the system automatically populates it with necessary properties used during scorecard execution.

You can learn more about creating and editing scorecards in the section.

data structure
dictionary
Scorecard Builder