Scribd
Upload
61 views3 pages

Lecture 7

The document discusses quantiles and percentiles. It defines quantiles as the inverse of the CDF for a continuous random variable X. The qth quantile is the value where the CDF equals q. Common quantiles include the median (50th percentile) and quartiles (25th, 50th, 75th percentiles). For discrete distributions, the quantile definition is modified to allow for intervals rather than single values. An example shows how to interpret a QQ plot by comparing the quantiles of two distributions. Order statistics are defined as the values of a random sample sorted from smallest to largest. The empirical CDF is constructed using the order statistics.

Uploaded by

amanmatharu22
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
61 views3 pages

Lecture 7

The document discusses quantiles and percentiles. It defines quantiles as the inverse of the CDF for a continuous random variable X. The qth quantile is the value where the CDF equals q. Common quantiles include the median (50th percentile) and quartiles (25th, 50th, 75th percentiles). For discrete distributions, the quantile definition is modified to allow for intervals rather than single values. An example shows how to interpret a QQ plot by comparing the quantiles of two distributions. Order statistics are defined as the values of a random sample sorted from smallest to largest. The empirical CDF is constructed using the order statistics.

Uploaded by

amanmatharu22
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Lecture - 7

Instructor: Dr. Arabin Kumar Dey

Quantitle/Percentile:

If the CDF of X is continuous and strictly increasing then it has an inverse function F 1 . For each q between 0 and 1, F 1 (q) is called the q-quantile or 100qth percentile. The probability that a continuous X is below its q-quantile is precisely q, but we will show this is not exactly true for discrete random variables.

The median is 50% percentile or 0.5-quantile. The 25% and 75% percentiles (0.25 and 0.75-quantiles) are called the rst and third quartiles and the median is second quartile. The three quartiles divide the range of the range of the random variable into four groups of equal probability. Similarly the 20%, 40%, 60% and 80% percentiles are called quintiles and 10%, 20%, , 90% percentiles are called deciles.

For discrete distributon we may see P (X < F 1 (q)) < q < P (X F 1 (q)). Also if the function is increasing but not strictly increasing, there exists an interval of q-quantiles. Therefore we modify the denition of quantile : The set q-quantile is the closed interval [x , x+ ] where: x = inf{x : F (x) q} and x+ = inf{x : F (x) > q}. q q q q We get a uniquely dened quantile if we have x = x+ . q q 1

Example 2.1. For exponential distribution qth qunatile q can be written as q = whereas for cauchy it can be written as q = m + tan(q ) 2 1 1 ln( ) 1q

Understanding the meaning of QQ plot

We are going to illustrate the intrepretation of QQ-plot with the experiment below as they relate to thickness of its tails. We calculate the quantiles for normal and cauchy distribution. Theoretical QQ-plot for these two distributions can be produced through by plotting quantiles of normal distribution against quantiles of cauchy distribution. X N (0, 1) 0.842 1.036 1.282 1.645 1.960 2.326 X C(0, 1) 1.376 1.963 3.078 6.314 12.706 31.821

1 0.8 = FX (0.8) 1 0.85 = FX (0.85) 1 0.9 = FX (0.9) 1 0.95 = FX (0.95) 1 0.975 = FX (0.975) 1 0.99 = FX (0.975)

In particular, we would have to plot the points (0.842, 1.376), (1.036, 1.963), (1.282, 3.078), , (2.326, 31.821). Note that all these points are above diagonal y = x, and in fact they drift further and further away above the diagonal. This fact is at the core of the interpretation of a QQ-plot comparing two distributions: points above the diagonal in the rightmost part of the plot indicate that the upper tail of the rst distribution (whose quantiles are on the horizontal axis) is thinner than the tail of the distribution whose quantiles are on the verticle axis. Similarly, points below the diagonal on the left part of the plot indicate that the second distribution has a heavier lower tail. 2

Order Statistics and Sample Quantile:

Suppose that X1 , X2 , , Xn is the random sample from a probability distribution F . We dene the empirical cdf or estimate of cdf from the sample as Fn (x) =
n i=1

I(Xi x) . n

where I(Xi x) is 1 if Xi x and 0 otherwise. We dene order statistics X(1) , X(2) , , X(n) are X1 , X2 , , Xn ordered from smallest to largest. X(1) is the smallest observation and X(n) is the largest observation. Suppose the sample is 6,4,8,2,3,4. Then the order statistics are 2,3,4,4,6,8. We write

Fn (x) =

0 1 6 2
6

4 6 5 6 1

if if if if if if

x<2 2x<3 3x<4 4x<6 6x<8 x>8

You might also like