4.1 What is Skewness?

Skewness is a fundamental statistical measure that quantifies the asymmetry of a dataset's probability distribution. In simpler terms, it tells us whether the data is evenly spread around its central tendency or if it's stretched more towards one side.

A perfectly symmetrical distribution (like a normal distribution) has its data evenly distributed on both sides of the center. In such a case, the mean, median, and mode are all equal.

However, real-world data rarely exhibits perfect symmetry. Skewness helps us identify whether the values in a dataset are stretched more on one side of the average than the other, providing critical insights into the shape and behavior of the data.

Types of Skewness

There are three primary types of skewness:

1. Positive Skewness (Right Skew)

In a positively skewed distribution, the right tail of the distribution is longer or fatter than the left tail. This indicates that there are a few unusually large values that pull the distribution's tail towards the right.

• Relationship between Measures: Mean > Median > Mode
• Characteristics:
  • The bulk of the data is concentrated on the left side.
  • A few extreme high values pull the mean upwards.
• Example: Income distribution is a classic example. Most people earn modest incomes, but a small number of individuals earn extremely high salaries, creating a long right tail.

2. Negative Skewness (Left Skew)

In a negatively skewed distribution, the left tail of the distribution is longer or fatter than the right tail. This suggests that there are a few unusually small values that pull the distribution's tail towards the left.

• Relationship between Measures: Mean < Median < Mode
• Characteristics:
  • The bulk of the data is concentrated on the right side.
  • A few extreme low values pull the mean downwards.
• Example: Exam scores can often exhibit negative skewness. Most students might perform well and score high marks, but a few students might struggle and achieve very low scores, creating a long left tail.

3. Zero Skewness (Symmetrical Distribution)

A distribution with zero skewness is perfectly symmetrical. This means the data is evenly distributed around the center.

• Relationship between Measures: Mean = Median = Mode
• Characteristics:
  • The data is balanced on both sides of the center.
  • The shape of the distribution is identical on both the left and right sides.
• Example: A normal distribution (bell curve) is the most common example of zero skewness.

Why Skewness Matters

Understanding skewness is crucial in various fields as it impacts data analysis and decision-making:

• Finance: A positively skewed return distribution might signal opportunities for substantial gains but also a higher risk of larger losses due to extreme values.
• Marketing and Customer Analytics: Skewness can highlight outliers or identify niche behaviors within a small segment of customers that might require specific attention or strategies.
• Business Decision-Making: Skewness influences the choice of appropriate statistical models and summary measures. For instance, using the mean might be misleading with highly skewed data, making the median a more robust choice.
• Data Science and Machine Learning: Many machine learning algorithms assume normally distributed data. Skewed data can lead to biased model performance, affecting accuracy and generalization. Techniques like data transformation are often employed to mitigate the impact of skewness.

Summary

Skewness is a vital statistic that describes the asymmetry of a dataset relative to its central tendency. It reveals how a dataset deviates from normality and highlights the presence of outliers or uneven data distribution. Recognizing whether a dataset is positively or negatively skewed enables more accurate data analysis, better model selection, and improved data-driven decisions.

---

Interview Questions

Here are some common interview questions related to skewness:

• What is skewness and why is it important in data analysis?
• How do you interpret positive and negative skewness?
• Can you explain the relationship between mean, median, and mode in skewed data?
• How does skewness affect statistical modeling and machine learning algorithms?
• What methods can be used to handle skewed data?
• How do outliers impact skewness?
• How is skewness calculated mathematically?
• Give an example of a real-world dataset that might be positively skewed.
• What is the difference between skewness and kurtosis?
• How can you visualize skewness in a dataset?