Templated AI Model Builder for Propensity Score Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current predictive engagement scores in cloud platforms fail to consider user profile data and are not configurable by end users, limiting their ability to predict user actions such as product purchases or churn, and thus do not provide a comprehensive view of customer behavior for targeted marketing.
Innovation Solution
A templated, no-code AI model builder is introduced that allows users to define prediction metrics for propensity scores, using both user engagement and profile data to generate custom scores for segmenting entities, enabling more accurate customer segmentation and targeted marketing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If predictive engagement scores are generated from external systems using only user engagement data, then the scoring process is simple, but the prediction accuracy and comprehensiveness of user behavior is insufficient
Solution Approach 1:
The patent merges multiple data sources (user profile data from data service and user engagement data from external systems) into a unified scoring mechanism. The propensity score generator integrates these diverse data types to create comprehensive predictions about user actions, resolving the contradiction by combining simple external scoring with comprehensive internal data to achieve higher prediction accuracy.
Solution Approach 2:
The propensity score generator is designed as a multi-functional component that handles both data collection from multiple sources, data processing, and prediction generation. This universal component can incorporate various data types (profile data, engagement data, contextual data) to serve multiple prediction needs, improving accuracy without proportionally increasing system complexity.
2Adaptability or versatility
If predictive engagement scores are generated from external systems, then the data service can focus on core functions, but the scores are not configurable by end users and fail to consider user profile data
Solution Approach 1:
The system provides dynamic configurability where end users can adjust propensity score parameters and data source selections based on their specific needs. The template-based approach allows users to dynamically configure which data sources to use, which prediction metrics to generate, and how to interpret results, adapting the system to different business requirements without requiring complex custom development.
Solution Approach 2:
The configuration capability is segmented into separate templates that users can select from or modify. Each template represents a configurable prediction model with specific parameters and data sources. This segmentation allows users to configure scores without managing complex system architecture, as they work with pre-defined templates rather than raw system complexity.
3Measurement precision
If comprehensive user profile data and engagement data are integrated for propensity scoring, then prediction accuracy improves, but data processing complexity and computational resources increase
Solution Approach 1:
The system performs preliminary data processing and feature engineering when data is first ingested, creating preprocessed data structures and cached predictions. This preliminary action reduces the computational burden during actual propensity score generation, as the system can leverage preprocessed data rather than processing raw data in real-time, thus improving accuracy while managing computational resources.
Solution Approach 2:
The system applies partial processing strategies by selectively processing only the necessary data and features required for each specific prediction task. Rather than processing all available data uniformly, the system identifies and processes only the relevant portions of user profile data and engagement data needed for each propensity score calculation, reducing overall computational resource consumption while maintaining prediction accuracy.
Data Source
AI summary
Methods, systems, apparatuses, devices, and computer program products are described. A data service may receive a first user input indicating a first set of entities for training an artificial intelligence (AI) model for propensity score-prediction. The data service may receive a second user input indicating a set of outcome conditions which define what a user would like to predict about a customer (e.g., propensity to purchase). The data service may generate the AI model accordingly, and based on executing the AI model, generate a set of prediction metrics (propensity scores) for a second set of entities. The data service may store an indication of the AI model for review by a user. When the user approves the AI model and publishes the AI model to the data service, the generated propensity scores may be used to generate a segment of entities of the second set of entities.