Explain the concept of skewness and kurtosis in Data science

In data science, skewness and kurtosis are two important statistical measures used to describe the shape and distribution of a dataset. They provide insights into the symmetry and the tails of the distribution, respectively. Here's a detailed explanation of each:

Skewness

Skewness measures the asymmetry of the probability distribution of a real-valued random variable about its mean. It indicates whether the distribution is skewed to the left or right of the mean.

1. Definition:

   • Formula: $$\text{Skewness} = \frac{n \sum_{i=1}^n (X_i - \bar{X})^3}{(n-1)(n-2) \sigma^3}$$where:
     • $n$ is the number of observations.
     • $X_i$ is each individual observation.
     • $\bar{X}$ is the sample mean.
     • $\sigma$ is the sample standard deviation.
   • Interpretation:
     • Positive Skewness: If skewness > 0, the distribution has a longer or fatter tail on the right side (right-skewed or positively skewed). The mean is greater than the median.
     • Negative Skewness: If skewness < 0, the distribution has a longer or fatter tail on the left side (left-skewed or negatively skewed). The mean is less than the median.
     • Zero Skewness: If skewness = 0, the distribution is perfectly symmetric, though this is rare in real-world data.

2. Example:

• If you have a dataset of income where most people earn relatively low amounts but a few earn very high incomes, the income distribution will be positively skewed. This is because the high-income outliers pull the tail of the distribution to the right.

Kurtosis

Kurtosis measures the "tailedness" of the probability distribution, indicating how heavily the tails of the distribution differ from the tails of a normal distribution.

1. Definition:

   • Formula: $$\text{Kurtosis} = \frac{n(n+1) \sum_{i=1}^n (X_i - \bar{X})^4}{(n-1)(n-2)(n-3) \sigma^4} - \frac{3(n-1)^2}{(n-2)(n-3)}$$ where:
     • $n$ is the number of observations.
     • $X_i$ is each individual observation.
     • $\bar{X}$ is the sample mean.
     • $\sigma$ is the sample standard deviation.
   • Interpretation:
     • Leptokurtic: Kurtosis > 3 (excess kurtosis > 0) indicates heavy tails or outliers; the distribution has more extreme values than a normal distribution (e.g., stock returns).
     • Platykurtic: Kurtosis < 3 (excess kurtosis < 0) indicates light tails or fewer outliers; the distribution has fewer extreme values than a normal distribution (e.g., uniform distribution).
     • Mesokurtic: Kurtosis = 3 (excess kurtosis = 0) indicates a normal distribution. This is often used as a baseline for comparison.

2. Example:

   • In financial markets, stock returns often exhibit leptokurtic behavior, meaning there are more frequent extreme returns than predicted by a normal distribution. This is useful for risk assessment and modeling.

Summary

• Skewness measures the asymmetry of a distribution. Positive skewness indicates a longer right tail, while negative skewness indicates a longer left tail. Zero skewness suggests a symmetric distribution.

• Kurtosis measures the "tailedness" of a distribution. High kurtosis (leptokurtic) indicates heavy tails with more outliers, while low kurtosis (platykurtic) indicates light tails with fewer outliers. Normal distributions have a kurtosis of 3, serving as a reference point.

Understanding skewness and kurtosis helps in assessing the shape of the distribution, which is crucial for selecting appropriate statistical methods and accurately interpreting the data.

https://mytecbooks.blogspot.com/2024/08/explain-concept-of-skewness-and.html

Explain the concept of skewness and kurtosis in Data science interview questions and answers interview questions answers freshers advanced experienced