Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 23 views

    Skewness and Kurtosis: An Example: Normal Lognormal

    Uploaded by

    hsathasivan
    Skewness and kurtosis describe the shape of a distribution. Skewness indicates asymmetry - a normal distribution is symmetric with zero skewness, while a log-normal distribution is right-skewed with positive skewness. Kurtosis measures outliers - a higher kurtosis means extreme values far from the center are more likely. A log-normal distribution has higher kurtosis than a normal distribution, as seen by its higher density in the right tail, indicating bigger values are more probable. Comparing normal and log-normal distributions with the same mean and variance demonstrates how skewness and kurtosis differ between symmetric and asymmetric, heavy-tailed distributions.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Skewness and Kurtosis: An Example: Normal Lognormal

    Uploaded by

    hsathasivan
    23 views2 pages
    Skewness and kurtosis describe the shape of a distribution. Skewness indicates asymmetry - a normal distribution is symmetric with zero skewness, while a log-normal distribution is right-skewed with positive skewness. Kurtosis measures outliers - a higher kurtosis means extreme values far from the center are more likely. A log-normal distribution has higher kurtosis than a normal distribution, as seen by its higher density in the right tail, indicating bigger values are more probable. Comparing normal and log-normal distributions with the same mean and variance demonstrates how skewness and kurtosis differ between symmetric and asymmetric, heavy-tailed distributions.

    Original Title

    Skewness Kurtosis

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Skewness and kurtosis describe the shape of a distribution. Skewness indicates asymmetry - a normal distribution is symmetric with zero skewness, while a log-normal distribution is right-skewed with positive skewness. Kurtosis measures outliers - a higher kurtosis means extreme values far from the center are more likely. A log-normal distribution has higher kurtosis than a normal distribution, as seen by its higher density in the right tail, indicating bigger values are more probable. Comparing normal and log-normal distributions with the same mean and variance demonstrates how skewness and kurtosis differ between symmetric and asymmetric, heavy-tailed distributions.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    23 views2 pages

    Skewness and Kurtosis: An Example: Normal Lognormal

    Uploaded by

    hsathasivan
    Skewness and kurtosis describe the shape of a distribution. Skewness indicates asymmetry - a normal distribution is symmetric with zero skewness, while a log-normal distribution is right-skewed with positive skewness. Kurtosis measures outliers - a higher kurtosis means extreme values far from the center are more likely. A log-normal distribution has higher kurtosis than a normal distribution, as seen by its higher density in the right tail, indicating bigger values are more probable. Comparing normal and log-normal distributions with the same mean and variance demonstrates how skewness and kurtosis differ between symmetric and asymmetric, heavy-tailed distributions.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    Skewness and Kurtosis: an Example

    Skewness

    To understand the notion of Skewness and Kurtosis, it is best to give an example. Let us consider two random
    variables:
    • Normal: has no Skeweness
    • Log-Normal: has positive Skewness (you can also say right-skewed)
    The Normal distribution is symmetric, hence is not skewed (has a Skewness of 0). In comparison, a
    distribution with positive Skewness (like the Log-Normal) is not symmetric and has a longer right tail
    than left tail. A negatively skewed random variable would have the opposite: a longer left tail. This is
    made clear on the densities plotted below (R code included for your reference).
    par(mar=c(3,6,1,1))
    x <- seq(-3, 7, by = 0.01)
    plot(x, dnorm(x, mean = 2, sd = 1), type = "l", col="red", lwd=4,
    xlab ="", ylab = "pdf", cex.lab = 2, cex.axis = 2, main = "", ylim = c(0, 0.55))
    curve(dlnorm(x, meanlog = log(2/sqrt(1.25)), sdlog = sqrt(log(1.25)) ), col="blue", lwd=4, add = TRUE)
    legend("topright", NULL, ncol=1, cex = 1.5, legend=c("Normal", "LogNormal"),
    col=c("red", "blue"), lty = c(1,1), lwd = c(5,5))

    Normal
    LogNormal
    0.4
    pdf
    0.2
    0.0

    −2 0 2 4 6
    For reference, note that both the Normal and Log-Normal displayed here have a mean of 2 and a variance of
    1. Hence, comparison is ‘fair’.

    1
    Kurtosis

    A high kurtosis means that ‘extreme values’ (to the left and/or right) are ‘somehow’ likely. What does
    ‘somehow’ mean? Well, more likely than for a distribution with small Kurtosis. What are ‘extreme’ values?
    Values far from the center of the distribution. Roughly, more than 3 standard deviations away can be
    considered far. Note that what the center of a distribution is is partly subjective: one could use the mean or
    the median (or even the mode) to define the center.
    Again we can use the Normal / Log-normal comparison. The Log-Normal has a higher Kurtosis than
    the Normal. This can be seen by the fact that the right tail is higher: meaning that ‘big values’ to the
    right are more likely (than for the Normal). This is more apparent if we zoom on those right tails.

    Normal
    0.04

    LogNormal
    pdf
    0.02 0.00

    4 5 6 7 8 9
    We see that for very big values (away from the center of the distribution), i.e. values of 5, 6, 7+, the density
    of the Log-normal is higher than that of the Normal. This illustrates the higher likelihood of extreme values
    to happen for the Log-normal (and hence the higher Kurtosis).

    You might also like