Its name may evoke mystical spells and 20-sided dice. In reality, MaGE is an EFT Analytics proprietary algorithm that empowers users to apply the knowledge of their process data by leveraging powerful machine learning algorithms.
Feels like magic. Works like science.
Most process engineers know a lot about their data, but not much about data science. So they may not know what data model best fits the problem they are trying to solve. Enter MaGE – to enhance our customers’ expertise and deep knowledge of the process and their data by thorough powerful, intuitive, easy-to-use machine learning algorithms.
How, you might ask? MaGE evaluates and picks the best model for you – saving you valuable time and delivering more insightful results.
How does MaGE do that?
The magic goes well beyond slight of hand. MaGE is an ensemble algorithm that uses multiple models to best learn the structure of the data. These “ensemble models” combine a diverse set of models to improve the stability and predictive power of the model. MaGE builds a variety of models in the platform as well as an ensemble model and provides scores for all of the models. These models include: Bayesian Network, Support Vector Machine, Decision Tree, Logistic Regression, Multilayer Perception Layer and Random Forest.
This process allows for increased stability and gives your models the predictive power of, dare we say, a wizard.
Step 1: Select a Data set
Step 2: Choose the MaGE model
Step 3: Configure your variables
- Select your Target Variable
- Select a primary target and/or sub-targets
Step 4: MaGE will score the models and select the best model to utilize
Step 5: Glean insights from the data you put into CORTEX™
Be on the lookout for MaGE Version 1.2 which incorporates a more powerful clustering algorithm behind the scenes.