Search Results ams_dm_lift




Overview
Explore in AI Chat

Overview

The AMS_DM_LIFT table is a core data repository within the Oracle E-Business Suite (EBS) Marketing (AMS) module, specifically for versions 12.1.1 and 12.2.2. It functions as a result table designed to store the output data generated by predictive modeling lift analysis. Lift analysis is a critical technique in marketing analytics used to measure the performance of a predictive model, such as a campaign response model, by comparing its results against a random selection. The data stored in this table is directly utilized for the generation of lift charts, which are visual tools that help marketers evaluate model effectiveness, optimize campaign targeting, and allocate resources efficiently by identifying the segments of the population most likely to respond.

Key Information Stored

The table's structure is engineered to capture the quantile-based results of a lift analysis for a specific predictive model. While the full column list is not detailed in the provided metadata, the primary and unique keys define its essential data points. The LIFT_ID column serves as the primary unique identifier for each lift record. The MODEL_ID column is a foreign key that links the lift data to its originating predictive model in the AMS_DM_MODELS_ALL_B table. The QUANTILE column indicates the percentile or segment (e.g., top 10%, 20%) for which the lift metrics are calculated. Typical data stored alongside these keys would include calculated metrics such as lift value, cumulative gain, response rate within the quantile, and possibly the number of respondents.

Common Use Cases and Queries

The primary use case is the creation and analysis of lift charts to validate and compare marketing models. Analysts run lift analysis for a deployed model, and the results populate this table. Common reporting queries involve joining to the model table to retrieve descriptive context. A typical analytical query would retrieve lift data for a specific model, ordered by quantile, to plot a lift chart:

  • SELECT quantile, lift_value, response_rate FROM ams_dm_lift WHERE model_id = :model_id ORDER BY quantile;

Another common pattern is comparing the lift performance of multiple models for the same campaign by joining results on the quantile. Data from this table may also be extracted into business intelligence tools for advanced visualization and historical performance tracking of marketing models.

Related Objects

The AMS_DM_LIFT table has a defined dependency within the AMS schema's data mining structure. As per the documented relationship data:

  • AMS_DM_MODELS_ALL_B: This is the primary related table. The foreign key AMS_DM_LIFT.MODEL_ID references this table, establishing that every lift analysis record is associated with a specific predictive model. This relationship is crucial for contextualizing lift results, allowing joins to retrieve model names, types, and creation details.

The table's integrity is enforced by its primary key constraint (AMS_DM_LIFT_PK on LIFT_ID) and a unique key constraint (AMS_DM_LIFT_UK1 on MODEL_ID and QUANTILE), the latter ensuring only one set of lift results exists per model per quantile.