Methods for Analyzing, Evaluating, and Exploring Models

After building a machine learning model in LityxIQ, you will likely want to analyze it.  LityxIQ provides a wide variety of methods evaluating and exploring models.  They are briefly explained here, with links to more detailed overviews of each.

Performance Analysis - analyze the quantitative and qualitative performance of the model, and compare models and versions against each other.  See for more.

Model Explorer - interactively explore the complex patterns and relationships among model variables.  See for more.

Threshold Analysis - interactively determine optimal cutoff values based on the cost of different kinds of prediction errors (only available for binary classification models).  See for more.

Engagement Analysis - interactively determine optimal depth for marketing campaigns to maximize profitability (only available for binary classification models).  see for more.

To get started with any of these, select the model you would like to evaluate or explore, and hover on Evaluate & Explore from the Selected Model menu or the right click menu.  This will open a sub-menu with the various options.