Histogram Plot

Purpose

Visualize the frequency distribution of a single continuous variable to understand its shape, central tendency, and spread.

Histogram = discrete distribution view
KDE = smooth distribution view

Analysis Type

Univariate

Documentation

• Matplotlib Histogram
• Seaborn Histogram

What to Look For

1. Distribution Shape

• Symmetric: Data evenly distributed around the center (potential normal distribution)
• Skewed Right (Positive Skew): Long tail on the right side
• Skewed Left (Negative Skew): Long tail on the left side
• Bimodal/Multimodal: Multiple peaks indicating subgroups

2. Normality Check

• Bell-shaped curve suggests normal distribution
• Important for many ML algorithms that assume normality

3. Outliers

• Values far from the main distribution
• Appear as isolated bars at extremes

4. Range and Spread

• Width of distribution indicates variability
• Narrow distribution = low variance
• Wide distribution = high variance

5. Linearity

• Linear:
  • Roughly symmetric or bell-shaped curve" (Normal) shape often pairs well with linear models (after standardization).
  • If both feature and target look fairly “well-behaved” (not extreme skew), linear relationships are easier to spot.
• Non-Linear:
  • A "Heavy Tail" or "Skewed" histogram usually indicates exponential or multiplicative behavior,thus the feature will have a non-linear relationship with other variables. They often suggest transformations that linearize relationships.
  • Multimodal histograms (two peaks) can signal hidden groups, which often creates non-linear patterns in scatter plots.

Code Example

# Basic histogram
plt.hist(df['column'], bins=30, edgecolor='black', alpha=0.7)
plt.title("Distribution of Variable")
plt.xlabel("Value")
plt.ylabel("Frequency")
plt.show()

# Seaborn with KDE overlay
sns.histplot(df['column'], kde=True, bins=30)
plt.title("Distribution with KDE")
plt.show()