Scoring Concepts: Network Generation & Design
Learn Effective Network Generation & Design Explore the essential concepts and tools behind network generation in Scoring Concepts: Network Generation & Design. Networks are central to Quantexa solutions, offering context to complex datasets through nodes and edges. Nodes can represent either: Documents, a hierarchical representation of raw data, or Entities, real-world concepts like individuals, businesses, or accounts created through Entity Resolution. This article covers: Network Design and Scoring: How to create networks that meet your project needs. Network Types: The differences between Natural Networks and Ego-centric Networks. Graph Script DSL Expansion Tips: Practical advice for generating high-quality ego-centric networks. For additional guidance, check out Graph Scripting Best Practices, or log in now to read the full article: 1. Scoring Concepts: Network Generation & Design - Quantexa Community This article introduces some concepts, tools and tips on the topic of network generation. By the end, you should understand which tool you would use to generate networks for your project and how to ensure the network output will cover the scores you have in your project. Introduction Networks are at the core of any…61Views0likes0CommentsScoring Concepts: Network Scoring
We can feed networks derived in Graph Scripting DSL into a Scoring pipeline to derive information and insight in the form of network-based scenarios. This article outlines: Concepts and approaches available for batch Network Scoring, including extracting information from graphs, testing, and debugging. Available tools and methods for performing network analysis with Assess. Read the article (login required): 2. Scoring Concepts: Network Scoring - Quantexa Community This article outlines the concepts and approaches available for batch Network Scoring, including extracting information from graphs, testing, and debugging. This article also highlights the available tools and methods for performing network analysis. The below steps are applicable for Assess only. Introduction We can feed…65Views0likes0CommentsScorecard Tuning📋
Scoring is a key part of the Quantexa Platform, and to maximise effectiveness needs to be tuned in an ongoing iterative process. There are various elements in the scoring process that can be tuned, and this article serves as a guide as to methods that can be employed to do so. Read the full article (login required): 1. Scorecard Tuning - Quantexa Community Introduction Scoring is a key part of the Quantexa Platform, and to maximise effectiveness, needs to be tuned in an ongoing iterative process. There are various elements in the scoring process that can be tuned, and this article will serve as a guide as to methods that can be employed to do so. What can be Tuned? There are…79Views1like0CommentsNew guide 📖 Scoring Concepts: Write-Once Steps
Discover how to efficiently use Write-Once Steps in the Assess framework for data transformation. This detailed guide complements the Write-Once Steps documentation and helps you determine when to apply Write-Once Steps effectively in both Batch and Dynamic Scoring contexts. Key topics covered: The potential cost of using the wrong method When to use Write-Once Steps vs. Logical Sources Strategies for scoring networks in Batch (SparkScoringContext) and Dynamic (DynamicScoringContext) environments Gain a deeper understanding of how to avoid duplicating logic across contexts and streamline your data engineering workflows. Read the full article (login required) to explore practical scenarios and best practices for scoring networks with Write-Once Steps: 5. Scoring Concepts: Write-Once Steps - Quantexa Community This article serves as an extension of the Product documentation of Write-Once Steps and provides a guide on which situation should the Write-Once Steps be used. Introduction The Write-Once Step is a data transformation step in the Assess framework, which is executable in both Batch and Dynamic contexts, meaning you do not…23Views0likes0CommentsScoring Concepts: Alerting and Re-alerting 📖
Alerting is the name for the process that comes after Scorecard creation and before Task loading into a Quantexa Deployment. Alerting decides which Subjects in the Scorecard output should alert to the end-users. Re-Alerting is an Alerting process that occurs after the first Alerting cycle, when new Scorecard data becomes available. Re-alerting logic compares a Subject's Scorecard output with all previous Scorecard outputs. The aim of this process is to ensure that there is new material risk that an Investigator would like to review. It is strongly encouraged that all deployments with Batch Scoring implement Alerting. Read about the Alerting Framework, Alerting Threshold, and Score Types in Scoring Concepts: Alerting and Re-alerting. Read the full article (login required): 4. Scoring Concepts: Alerting and Re-alerting - Quantexa Community This article builds upon the concepts introduced in Scoring Concepts: Scoring Levels and Scorecards. Alerting is the name for the process that comes after Scorecard creation and before Task loading into a Quantexa Deployment. Alerting decides which Subjects in the Scorecard output should alert to the end-users. Re-Alerting…135Views1like0CommentsIndividual Score Tuning Guidance
Tuning happens at different levels of granularity and in the Quantexa scoring model the lowest level of granularity is the individual Score. One or more Scores contribute to a Scorecard and the combined contributions give a total Scorecard value for alerting. The objective of tuning a Score is to show that it identifies intended behavior suitably with the configurations given by business. The recommended approach in Quantexa is above the line (ATL) and below the line (BTL) validation. Guidance for the first stage of tuning which is the individual Score level along with detailed steps of a synthetic worked example are covered in Individual Score Tuning Guidance (login required). Individual Score Tuning Guidance - Quantexa Community Tuning happens at different levels of granularity and in the Quantexa scoring model the lowest level of granularity is the individual Score. One or more Scores contribute to a Scorecard and the combined contributions give a total Scorecard value for alerting. The following guidance will be for the first stage of tuning…150Views1like0Comments