On-Premise Predictive Model Training via Segmented Local Computation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Business teams lack expertise in data science, statistics, and programming to build predictive models, leading to challenges in increasing revenue, reducing revenue leakage, and prioritizing customer satisfaction, and face issues with data security and compliance when using third-party cloud services for machine learning.
Innovation Solution
A desktop application with a user interface allows users to build custom predictive models without prior knowledge of data science or machine learning, performing feature engineering, data cleansing, and model training, while ensuring data integrity and compliance by processing computations internally and using encryption for secure authentication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If business teams use third-party cloud services for machine learning, then they can access advanced predictive modeling capabilities, but they face data security and compliance risks
Solution Approach 1:
The patent introduces an on-premises machine learning environment as an intermediary solution between business teams and predictive modeling capabilities. This local deployment allows teams to access advanced ML functions without transmitting sensitive data to external cloud services, thus maintaining security while enabling adaptability. The system acts as a mediator that provides cloud-like capabilities in a secure, local context.
2Adaptability or versatility
If business teams build custom predictive models, then they can address specific functional problems, but they require expertise in data science, statistics, and programming
Solution Approach 1:
The patent implements automated model selection and training functionality that allows business teams to create custom predictive models without requiring data science expertise. The system automatically selects appropriate algorithms, performs feature engineering, and trains models based on user-defined parameters. This self-service approach enables non-experts to build customized models by simply specifying their business requirements, eliminating the need for deep technical knowledge while maintaining model customization capability.
3Reliability
If the system processes computations locally, then data security and compliance are maintained, but processing power and memory resources are limited
Solution Approach 1:
The patent divides the machine learning workload into segments that can be processed locally within resource constraints. The system breaks down complex modeling tasks into smaller, manageable computation units that fit within available memory and processing capacity. This segmentation allows the system to perform sophisticated analytics locally without requiring excessive resources, maintaining data compliance while achieving practical processing capability through distributed, modular computation.
Data Source
AI summary
The disclosure relates to a method for generating a custom predictive model, the method comprising receiving a plurality of datasets; identifying a plurality of features affecting prediction of a predictive model; determining an importance score for each of the plurality of features; determining a probability of prediction for each dataset from the plurality of the datasets based on one or more features from the plurality of features for the prediction of each dataset, and respective importance scores for the one or more features; and training a custom predictive model using the plurality of datasets with the probabilities, the respective features, and the respective importance scores.