1.
Q:- Describe non parametric test for understanding two means
Ans:- A non-parametric test for comparing two means is used when the assumptions of
a parametric test (such as normality and equal variances) are not met. These tests do not
rely on specific distribution assumptions and are useful for small sample sizes or ordinal
data. If data is not normally distributed and you have independent or dependent
samples then use Mann-Whitney U Test or Wilcoxon Signed-Rank Test.
Non-Parametric Tests for Comparing Two Means:
1. Mann-Whitney U Test
o Used when two independent groups need to be compared.
o It ranks all data points from both groups together and compares the sum of ranks.
o Suitable for ordinal or continuous data that is not normally distributed.
2. Wilcoxon Signed-Rank Test
o Used when comparing two related (paired) groups, such as pre-test vs. post-test
measurements.
o It considers the differences between paired observations and ranks them.
o Suitable for non-normal distributions in paired data.
2. Q:- Describe independent t test and paired sample t test with suitable example.
OR
Q:- Describe paired and non paired t test
1. Independent t-Test (Unpaired t-Test)
1. Compares the means of two independent groups.
2. Used when groups are not related (e.g., different participants in each group).
3. Assumes normality and equal variances (though Welch’s t-test can be used if variances
are unequal).
For example, A researcher wants to compare the average blood pressure between two
groups:
Group A: Patients given Drug A
Group B: Patients given Drug B
The test checks if the mean blood pressure differs between the two independent groups.
X 1−X 2
t=
√
2 2
s1 s2
+
n1 n 2
2. Paired Sample t-Test (Dependent t-Test)
1. Compares the means of two related groups (before-after scenario).
2. Used when the same subjects are measured twice (e.g., pre-test vs. post-test).
3. Assumes normality of the differences between paired observations.
D
t=
S. D
√n
Example:
For example, A researcher tests whether a speech therapy program improves pronunciation
skills:
Measure pronunciation before therapy (Pre-test)
Measure pronunciation after therapy (Post-test)
The test checks if the mean pronunciation score changed significantly.
Differences between Independent t- Test and Paired t-Test
Feature Independent t-Test Paired Sample t-Test
Relationship Groups are independent Groups are related (paired)
Sample Type Different individuals in each group Same individuals measured twice
Comparing drug effects between two Comparing test scores before and after
Example
groups training
Data
Two independent samples Paired observations
Requirement
4. Q:- Describe normal and non normal distribution of data. Also write the different variant
from the normal distribution
Ans:- Normal Distribution- A normal distribution (also called a Gaussian distribution or bell
curve) is a symmetrical, unimodal, and continuous probability distribution where most values
cluster around the mean. For example height of individuals in a population, IQ scores, Blood
pressure measurements
1. Bell-shaped curve: Symmetrical around the mean (μ\muμ).
2. Mean = Median = Mode.
3. 68-95-99.7 Rule (Empirical Rule):
a. 68% of data falls within 1 standard deviation (σ\sigmaσ).
b. 95% of data falls within 2 standard deviations.
c. 99.7% of data falls within 3 standard deviations.
4. Used in parametric tests like the t-test, ANOVA, and regression.
Non-Normal Distribution-A non-normal distribution does not follow the bell-shaped curve
and may have skewness, multiple peaks, or heavy tails.
Types of Non-Normal Distributions:
1. Skewed Distribution
o Left-skewed (Negative skew): Tail is longer on the left (e.g., age at retirement).
o Right-skewed (Positive skew): Tail is longer on the right (e.g., income
distribution).
2. Bimodal Distribution
o Two peaks instead of one (e.g., test scores of two different student groups).
3. Uniform Distribution
o All values occur with equal frequency (e.g., rolling a fair die).
4. Exponential Distribution
o Skewed right, used for modeling time between events (e.g., time between AI
system failures).
5. Heavy-Tailed Distribution (Laplacian or Cauchy)
o More extreme values than a normal distribution (e.g., stock market returns).
Variants of Normal Distribution
Different distributions arise from modifications to the normal distribution:
Variant Description Example
Standard Normal Normal distribution with mean = 0 and
Z-scores in statistics
Distribution standard deviation = 1
Multivariate Normal Generalization of normal distribution to Machine learning models,
Distribution multiple variables PCA
Log-Normal Log-transformed values follow a normal
Stock prices, biological data
Distribution distribution
Truncated Normal Height distribution in a
Normal distribution with cut-off limits
Distribution military selection
Differences: Normal vs. Non-Normal Distribution
Feature Normal Distribution Non-Normal Distribution
Shape Bell-shaped, symmetric Skewed, multimodal, heavy-tailed
Mean, Median,
Equal May differ
Mode
Example Height, IQ Income, stock market returns
Parametric tests (t-test, Non-parametric tests (Mann-Whitney U,
Statistical Tests
ANOVA) Kruskal-Wallis)