Statistics for Data Science

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.
Topics covered so far
1. Statistical Inference
a. Distributions - Binomial, Uniform, Normal
b. Sampling
c. Central Limit Theorem
d. Confidence Intervals
2. Hypothesis Testing
a. Hypothesis Formulation
b. One-Tailed Test vs Two-Tailed Test
c. Type I and Type II Errors

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 2
Gauge your Understanding
1. What is a random variable and how is it related to probability distribution?
2. What are some of the most commonly used distributions?
3. What is Central Limit Theorem (CLT) and when is it used?
4. What do you mean by estimations?

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 3
What is a Random Variable?
A random variable is a function that assigns a numerical value to each outcome of an experiment. It
assumes different values with different probability. It is usually denoted by capital letter X and the
probability associated with any particular value of X is denoted by P(X=x).

Example: Suppose that a fair coin is tossed twice and the possible outcome are {HH, HT, TH, TT}. Let X be
the random variable representing the number of heads that can come up. So, X can take values from the
set {2, 1, 0}.
The probability of two heads coming up is P(X=2) = ¼.

Random Variable

Discrete random variable: It can Continuous random variable: It can

take only a finite number of values. take uncountable number of values
For example: Number of employees in a given range. For example:
getting promoted in an organization. Speed of an aircraft.

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 4
What is a Probability Distribution?
The probability distribution of a random variable describes the values that the random variable can
take along with the probabilities of those values.

Discrete
Discrete Probability Probability mass
Distribution function

provides the probability for each

value of the random variable
Random Variable

Continuous Probability Probability density

Continuous
Distribution function

determines the probability with

which the continuous random
variable lies in a given interval

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 5
Distributions around us (commonly occuring)

Bernoulli The outcome of tossing a fair coin

Binomial The number of non-defective products in a production run

Uniform The number of books sold weekly at a bookstore

Normal IQ distribution of all the seven years old children in New York

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 6
Binomial Distribution
The binomial distribution is the probability distribution of the number of successes of an experiment that is
conducted multiple times and has only two possible outcomes.

Example: Suppose you have purchased 10 lottery tickets and the possible outcomes are winning the lottery or not
winning the lottery, then you can answer a question like what is the probability of winning 6 lottery tickets using
binomial distribution.

The assumptions of Binomial distribution are as follows:

1. There are only two possible outcomes (success or failure) for each trial.
2. The number of trials is fixed.
3. The outcome of each trial is independent. In other words, none of the trials have an effect on the probability
of the next trial.
4. The probability of success is exactly the same for each trial.

Note: In binomial distribution, if the number of trials for a given experiment is equal to 1, then it is called Bernoulli
distribution.

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 7
Uniform Distribution

The Uniform Distribution is the probability

distribution where all outcome are equal
likely.

Continuous Uniform Distribution: Can take

Discrete Uniform Distribution: Can take a any value within a given range with equal
finite number (m) of values and each value probability.
has equal probability of selection.
Example: Weight gained by a person over
Example: Rolling a single die. next 2 months can be uniformly distributed
between 2 to 5 Kg.

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 8
Normal Distribution
The normal distribution is a continuous probability distribution that is symmetric about the mean. It is also known
as bell curve because the graph of its probability density function looks like a bell.

Example: The height of all adult males in a city

Properties:

● It has a zero skewness

● Mean = Median = Mode
● If mean = 0 and standard deviation = 1, then it is called a standard normal distribution

Empirical Rule

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 9
Sampling Distributions
What is the need for sampling?

Given the limited resources and time, it is not always possible to study the population. That’s why we choose a sample out of the
population to make inference about the population.

Example: Suppose a new drug is manufactured and it needs is to be tested for the adverse side effects on a country’s population. It is
almost impossible to conduct a research study that involves everyone.

What are Sampling Distributions?

It is a distribution of a particular sample statistic obtained from all possible samples drawn from a specific population.

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 10
Central Limit Theorem

The sampling distribution of the sample means will approach

normal distribution as the sample size gets bigger, no matter
what the shape of the population distribution is.

Assumptions

Data must be randomly sampled Sample values must be independent of each other

Samples should come from the same distribution Sample size must be sufficiently large (≥30)

Let’s see CLT in action by simulation - Link to external site

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 11
Estimations

Make inferences about population

Estimation parameter based on sample statistic

Point Estimation Interval Estimation

The range of values within

which the population
Single value of a
parameter lies with some
population parameter
confidence
Ex. Population mean as
Ex. Population mean
estimated from the
should lie in the range
sample mean is $20
$15-$25, with 95%
confidence

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 12
 Case Study
Inferential Statistics

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 13
Gauge Your Understanding
1. What is hypothesis testing and what are different types of hypotheses?
2. What are some of the key terms involved in hypothesis testing?
3. What is the difference between one-tailed and two-tailed tests?
4. What are the steps to perform a hypothesis test?

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 14
Introduction to Hypothesis Testing

Question of Interest Hypotheses about the population

e.g. Has the new online parameter(s)
Ad increased the
conversion rates for an
E-commerce website?

Null Hypothesis (H0) Alternative Hypothesis (Ha)

The status quo The research hypothesis
e.g. The new Ad has not increased the e.g. The new Ad has increased the
conversion rate. conversion rate.

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 15
Key terms in Hypothesis Testing

● Probability of observing equal or more extreme results than the computed

test statistic, under the null hypothesis.
P-Value ● The smaller the p-value, the stronger the evidence against the null
hypothesis.

● The significance level (denoted by α), is the probability of rejecting the null
hypothesis when it is true.
Level of Significance ● It is a measure of the strength of the evidence that must be present in the
sample data to reject the null hypothesis.

● The total area under the distribution curve of the test statistic is partitioned
Acceptance or Rejection into acceptance and rejection region
Region ● Reject the null hypothesis when the test statistic lies in the rejection region,
else we fail to reject it

Types of Error ● There are two types of errors - Type I and Type II

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 16
Type I and Type II errors

Level of significance =
α
Confidence Level = H0 True H0 False
(1 - α )

Correct
Reject H0 Type I Error (α)
decision

Fail to reject Correct Type II Error

H0 decision (β)

Power of the
test = (1 - β)

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 17
Let’s go through an example
Problem Statement: The store manager believes that the average waiting time for the customers
at checkouts has become worse than 15 minutes. Formulate the Null and the Alternate
hypotheses.

Null Hypothesis (H0): The average waiting time at checkouts is less than equal to 15 minutes.

Alternate Hypothesis (Ha): The average waiting time at checkouts is more than 15 minutes.

Type I error (False Positive): Reject Null hypothesis when it is indeed true. “The fact is that the
average waiting time at checkout is less than equal to 15 minutes but the store manager has
identified that it is more than 15 minutes”.

Type II error (False Negative): Fail to reject Null hypothesis when it is indeed false. “The fact is
that the average waiting time at checkout is more than 15 minutes but the store manager has
identified that it is less than equal to 15 minutes”.

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 18
One-tailed vs Two-tailed Test

Reject H0 if the value of

Reject H0 if the value of Reject H0 if the value of
test statistic is either too
test statistic is too small test statistic is too large
small or too large

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 19
Hypothesis Testing Steps
Formulate H0 and Ha

Select Appropriate Test

Set Level of Significance, 𝛂

Collect Data and Calculate Test Statistic

Determine p-value Determine Critical Value

Compare with 𝛂 Compare with Test Statistic

Reject or Fail to Reject H0

Draw Conclusion
Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 20
 Case Study
Hypothesis Testing

Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited. 21
Happy Learning !

22
Proprietary content. © Great Learning. All Rights Reserved. Unauthorized use or distribution prohibited.