Conversion Prediction Model Feature Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Businesses face challenges in targeting and acquiring customers effectively, as identifying relevant data sources and determining which prospects are most likely to convert is complex due to the multitude of variables involved in the conversion process.
Innovation Solution
The method involves receiving historical data, performing correlation shape analysis to select highly correlated input variables, training machine learning models, and generating a feature set to predict conversion likelihood and other aspects of the conversion process, such as time and cost, to create look-a-like prospects and conversion prediction models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If hundreds of data sources and variables are combined to improve conversion prediction accuracy, then prediction reliability improves, but system complexity increases
Solution Approach 1:
The patent segments the conversion prediction process into distinct modules: data collection from multiple sources, feature engineering, model training, and prediction generation. This segmentation allows complex data from hundreds of sources to be systematically processed through standardized stages, reducing overall system complexity while maintaining prediction accuracy.
Solution Approach 2:
The patent introduces an intermediary feature engineering layer that transforms raw data from multiple sources into standardized features. This intermediary layer acts as a mediator between the complex data sources and the prediction model, simplifying the data integration process and making the system more manageable while preserving predictive power.
2Ease of operation
If automatic feature selection is implemented to reduce variables from thousands to manageable set, then ease of operation improves, but information loss may occur
Solution Approach 1:
The patent implements self-service feature selection through automated machine learning algorithms that automatically identify and select relevant features from thousands of variables. The system performs correlation analysis, identifies significant patterns, and selects optimal features without manual intervention, making the process easy to operate while using multiple selection criteria to minimize information loss.
Solution Approach 2:
The patent applies parameter changes in the feature selection process by transforming raw data into standardized features and adjusting selection criteria based on statistical significance and predictive power. This allows the system to automatically filter variables while preserving essential information through rigorous parameter-based selection thresholds.
3Productivity
If machine learning models are trained on historical data to predict conversion aspects, then productivity improves, but computational resources increase
Solution Approach 1:
The patent performs preliminary action by pre-processing and engineering features from historical data before model training. This preliminary feature engineering reduces the dimensionality and complexity of the data, allowing more efficient training with fewer computational resources while maintaining high productivity in customer acquisition predictions.
Solution Approach 2:
The patent applies partial action by training models on a strategically selected subset of the most informative features rather than all available data. This partial approach to data processing reduces computational resource consumption while maintaining sufficient predictive accuracy to drive efficient customer acquisition strategies.
Data Source
AI summary
Methods, systems and apparatuses, including computer programs encoded on computer storage media, are provided for generating prediction models related to targeting and acquiring customers. Thousands of variables of historical data, including data for prospects and external data, are used to train the prediction models. The variables are pre-processed, then sensitivity analysis is performed on the input variables with respect to the target. The variables with the most influence on the target are selected and added to the feature set used for training a prediction model.