Distributions & Outliers

A mean of 52 tells you almost nothing about an age column. The distribution — the full shape of the values — is what matters. Is it a bell curve around 52? Two humps at 25 and 70? A suspicious spike at 0?

Plot a histogram first, always:

• Normal-ish: a symmetric hump — mean and standard deviation describe it well
• Long-tailed: a few huge values stretch the right side and drag the mean toward them
• Bimodal: two peaks often mean two hidden groups — hospital wait times mixing ER arrivals and scheduled visits
• Spiky: sharp spikes on round numbers or default dates — often data-entry placeholders, not real values