📊

Histogram Shape Analysis

Jul 12, 2025

Overview

This lecture focused on analyzing the shape of histograms for numerical variables by examining symmetry/skewness, the number of modes, and the presence of outliers.

Analyzing Histogram Shapes

  • Histograms are used to graph numerical variables such as number of visits or birth weight.
  • Three main questions to ask about histogram shape: symmetry/skewness, number of modes, and presence of outliers.
  • Symmetry can be symmetric, skewed left (left tail), or skewed right (right tail).
  • Mode refers to the number of distinct peaks: unimodal (one), bimodal (two), or multimodal (more than two).
  • Outliers are data points that are far from the rest of the distribution.

Example: Left Graph (Number of Visits)

  • Shape: Skewed right because the shortest bars are on the right.
  • Mode: Unimodal since only one clear peak is present.
  • Outliers: Obvious outliers in the right tail.

Example: Right Graph

  • Shape: Skewed left due to a tail on the left.
  • Mode: Unimodal since there is only one peak in the outline.
  • Outliers: Not obvious; the graph is inconclusive for outliers.

Clarifying Modes

  • Unimodal means one peak, based on the outline, not individual bar numbers.
  • Bimodal requires two distinct peaks higher than the surrounding bars, not necessarily of equal height.
  • Multiple tall bars together do not automatically mean bimodal if the outline forms just one peak.

Key Terms & Definitions

  • Histogram — bar graph showing the distribution of a numerical variable.
  • Skewness — describes if the histogram tail is longer on the right (right-skewed) or left (left-skewed).
  • Mode — the number of peaks in a distribution’s outline (unimodal: one, bimodal: two, multimodal: more).
  • Outlier — data point significantly distant from the rest of the distribution.

Action Items / Next Steps

  • Learn the mathematical rule for defining outliers in section 3.5 of the course.
  • Practice identifying shape, mode, and outliers in more histogram examples.