Feature Analysis Controls

Overview

Feature Analysis in Mantis provides tools and visualizations for understanding the significance, distribution, and relationships of features within your space. It helps you identify informative variables, outliers, and trends, guiding deeper exploration and insight generation.

Purpose

• Assess the distribution and relevance of individual features across your data.
• Compare features to uncover trends, patterns, or discriminating variables.
• Detect outliers, correlations, or anomalies within and between groups.
• Support hypothesis testing and interpretation of embedding spaces.
• Inform downstream steps such as clustering, bag creation, or plot configuration.

Key Features

• Summary Statistics: Instant calculation of mean, median, mode, range, and variance for selected features.
• Visualization Tools: Integrated histograms, boxplots, and scatterplots for feature exploration.
• Correlation Analysis: Tools to examine relationships between pairs or groups of features.
• Group Comparison: Compare feature values across bags, clusters, or selections.
• Agent Integration: Summarize or interpret feature relevance using agent-driven analysis.
• Outlier Detection: Highlight data points that deviate significantly from feature distributions.

Tips

• Use feature analysis early to focus your exploration on the most informative variables.
• Compare features between bags or clusters to reveal underlying structure or drivers.
• Run agent-powered analysis to surface key insights or automate exploratory steps.