Raunak Yadush

Bhavesh Arora
Crack Top Analyst Roles with This 15-
Day STATISTICS Series!

Learn Complete STATISTICS

with Most ASKED INTERVIEW
Questions
Are you aiming for roles like Business Analyst, Data Analyst, Power BI
Developer, or BI Consultant at top companies?

Curated by
SAURABH G
Founder at DataNiti
6+ Years of Experience | Senior Data Engineer
Linkedin: www.linkedin.com/in/saurabhgghatnekar

BHAVESH ARORA
Senior Data Analyst at Delight Learning Services
M.Tech – IIT Jodhpur | 3+ Years of Experience
Linkedin: www.linkedin.com/in/bhavesh-arora-11b0a319b
Connect with us: https://topmate.io/bhavesh_arora/

Let’s embark on this journey together and

make your dreams a reality, starting
today.
🔹 Day 1: Descriptive Statistics – Mean, Median, Mode,
Range
Descriptive statistics summarize and describe the main features of a
dataset. These measures help in understanding data distribution.

✅ 1. Mean (Average)
The sum of all values divided by the number of values.

Python Code:
import numpy as np
data = [10, 20, 30, 40, 50]
print("Mean:", np.mean(data))

✅ 2. Median
The middle value when data is sorted.
● If odd: middle value
● If even: average of two middle values
Python Code:
print("Median:", np.median(data))

✅ 3. Mode
The most frequently occurring value in a dataset.
Python Code:
from scipy import stats
print("Mode:", stats.mode(data).mode[0])
✅ 4. Range
The difference between the highest and lowest value.
Formula:

Python Code:
print("Range:", max(data) - min(data))

💬 INTERVIEW QUESTIONS (Medium to High)

🔸 Q1: When would you prefer median over mean?
✅ When the data has outliers or is skewed, median gives a better
central tendency.

🔸 Q2: Can a dataset have more than one mode?

✅ Yes — it's called bimodal or multimodal depending on the number
of modes.

🔸 Q3: Why is range not always a reliable measure of dispersion?

✅ Because it's highly affected by outliers; it only considers the
extremes.

🔸 Q4: In what scenarios is the mean misleading?

✅ In skewed distributions (e.g., income, house prices), mean doesn’t
reflect the real central value.
🔹 Day 2: Measures of Dispersion – Variance, Standard
Deviation, IQR

Understanding how spread out your data is can be just as important as

knowing the average.

✅ 1. Variance (σ² or s²)

Variance tells how far each number in the dataset is from the mean.

Formulas:
● Population Variance (σ²):

Python Code:

import numpy as np
data = [10, 20, 30, 40, 50]
print("Sample Variance:", np.var(data, ddof=1)) # ddof=1 for sample

✅ 2. Standard Deviation (σ or s)
The square root of variance. It’s in the same unit as the data.
Formula:
Python Code:
print("Standard Deviation:", np.std(data, ddof=1))

✅ 3. Interquartile Range (IQR)

IQR measures the middle 50% spread of the data.

Formula:

Python Code:
Q1 = np.percentile(data, 25)
Q3 = np.percentile(data, 75)
print("IQR:", Q3 - Q1)

💬 INTERVIEW QUESTIONS (Medium to High)

🔸 Q1: What is the relationship between variance and standard
deviation?
✅ Standard deviation is the square root of variance, making it more
interpretable in real-world terms.
🔸 Q2: Why do we divide by (n - 1) in sample variance?
✅ To correct the bias in estimation — this is called Bessel’s correction.
🔸 Q3: Why is IQR preferred over range in some cases?
✅ IQR ignores outliers, making it more robust for skewed
distributions.

🔸 Q4: When is a high standard deviation a concern?

✅ When consistency is desired (e.g., product quality, risk in
investments), a high SD means high variability.
🔹 Day 3: Probability Basics – Types, Rules & Real-
World Use Cases
Probability forms the foundation of statistics and is critical in data
analysis, risk modeling, and machine learning.

✅ 1. Types of Probability
Type Description Example
Theoretical Based on reasoning or Coin flip: P(Heads) =
known outcomes 0.5
Based on actual
experiments/data 3 out of 10 students
Based on intuition or
Experimental are left-handed →
experience P(LH) = 0.3
“I feel there’s a 70%
Subjective chance of rain”

✅ 2. Basic Probability Formula

Python Code:
favorable = 3
total = 10
print("Probability:", favorable / total)
✅ 5. Real-World Use Cases

● Spam filters use conditional probability (Bayes’ Theorem)

● Risk analysis in insurance and banking
● Predictive modeling in ML algorithms
●
A/B testing and marketing success prediction
💬 INTERVIEW QUESTIONS (Medium to High)
🔸 Q1: What’s the difference between independent and mutually
exclusive events?
✅ Independent: Events don’t influence each other.
✅ Mutually Exclusive: Both events cannot occur together.

🔹 Day 4: Conditional Probability & Bayes’ Theorem –

Intuition to Application

Conditional probability and Bayes’ Theorem are the backbone of

predictive analytics, machine learning, and risk assessment.
✅ 3. Python Code Example
# Bayes' Theorem Calculation
P_A = 0.01 # Prior: Disease
P_B_given_A = 0.99 # Test positive if disease
P_B_given_not_A = 0.05 # False positive
P_not_A = 1 - P_A

P_B = (P_B_given_A * P_A) + (P_B_given_not_A * P_not_A)

P_A_given_B = (P_B_given_A * P_A) / P_B
print("P(Disease | Positive):", P_A_given_B)

✅ 4. Use Cases of Bayes' Theorem

● Spam filters (Gmail, Outlook)
● Medical diagnostics
● Recommendation systems
● Credit scoring & fraud detection

💬 INTERVIEW QUESTIONS (Medium to High)

🔸 Q1: What’s the difference between P(A|B) and P(B|A)?
✅ They are not the same! Bayes' Theorem connects them.
🔸 Q2: Why is Bayes’ Theorem important in ML?
✅BayesIt allows updating predictions based on new evidence (e.g., Naive
Classifier).

🔸 Q3: What’s the biggest trap in interpreting conditional probability?

✅ Ignoring the base rate (prior probability). It can skew results
drastically.

🔸 Q4: How would you explain conditional probability to a non-

technical person?
✅ “If it’s raining, the chance of traffic increases. That’s conditional
probability—knowing one thing changes the likelihood of another.”
🔹 Day 5: Probability Distributions – Discrete vs
Continuous
Probability distributions tell us how likely outcomes are—the foundation
for modeling uncertainty in real-world data.
✅ 1. Discrete Distributions
These apply when outcomes are countable (whole numbers).
📌 Common Examples:
● Bernoulli Distribution (Single binary trial: Success/Failure)
● Binomial Distribution (Multiple independent binary trials)
● Poisson Distribution (Number of events in a fixed interval)
✅ 3. Python Code Example
import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import binom, norm

# Binomial (n=10, p=0.5)

x = np.arange(0, 11)
binom_probs = binom.pmf(x, 10, 0.5)
plt.bar(x, binom_probs)
plt.title('Binomial Distribution (n=10, p=0.5)')
plt.show()
# Normal
x = np.linspace(-3, 3, 100)
plt.plot(x, norm.pdf(x, 0, 1))
plt.title('Normal Distribution (μ=0, σ=1)')
plt.show()
✅ 4. When to Use Which
Scenario
Use
Count of events (e.g., calls per hour)
Repeated binary trials Poisson
Heights, weights Binomial
Waiting times Normal
Exponential

💬 INTERVIEW QUESTIONS (Medium to High)

🔸 Q1: What’s the difference between discrete and continuous
variables?
✅ Discrete = Countable, Continuous = Measurable
🔸 Q2: How is the binomial distribution related to Bernoulli?
✅ Binomial is a sum of independent Bernoulli trials
Q3: Why is the normal distribution so widely used?
🔸 Central Limit Theorem: Means of samples from any distribution
✅become normal as sample size grows.
🔸 Q4: Can probabilities be assigned to exact values in continuous
distributions?
✅ No. For continuous variables, P(X = x) = 0. Only ranges matter (e.g.,
P(2 < X < 5)).

🔹 Day 6: Central Limit Theorem (CLT) – Why the Bell

Curve Appears Everywhere

The Central Limit Theorem (CLT) is the bedrock of statistical inference

and hypothesis testing.
✅ 1. What is the Central Limit Theorem?
When you take many random samples from any population (regardless
of its shape), the distribution of sample means will:
● Tend toward a Normal Distribution
● Have a mean = population mean (μ)
● Have a standard deviation = σ / √n (called Standard Error)
As n (sample size) increases, normality improves.

✅ 2. Why It’s Powerful

Even if your population is:
● Skewed
● Bimodal
● Non-normal

👉 The mean of large enough samples will still follow a normal

distribution.
This allows:
● Hypothesis testing
● Confidence intervals
● Z-tests

✅ 3. Visualize CLT in Python

import numpy as np
import matplotlib.pyplot as plt
population = np.random.exponential(scale=2, size=10000)
sample_means = []
for _ in range(1000):

sample = np.random.choice(population, size=50)

sample_means.append(np.mean(sample))

plt.hist(sample_means, bins=30, density=True)

plt.title("Sampling Distribution of the Mean")
plt.xlabel("Sample Mean")
plt.ylabel("Frequency")
plt.grid(True)
plt.show()
💡 Even though the population is exponential (skewed), the sample
means form a bell curve!
🔹 Day 7: Hypothesis Testing – From Assumptions to Conclusions
Hypothesis Testing helps you make decisions based on data — whether
you're testing marketing strategies, A/B experiments, or scientific
claims.

✅ 1. What is Hypothesis Testing?

It’s a statistical method to test an assumption (claim) about a population
using sample data.
● Null Hypothesis (H₀): Status quo / no effect
● Alternative Hypothesis (H₁): Your claim / there is an effect
Example:
H₀: The average delivery time = 2 days
H₁: The average delivery time ≠ 2 days
✅ 2. Steps in Hypothesis Testing
1. Define H₀ and H₁
2. Select significance level (α) – typically 0.05
3. Choose test type (z-test, t-test, etc.)
4. Calculate test statistic
5. Compare with critical value or p-value
6. Make decision: Reject or fail to reject H₀

✅ 3. Types of Errors
Type Meaning
Type I Error Rejecting H₀ when it's actually true (False Positive)
Type II Error Failing to reject H₀ when it's false (False Negative)

✅ 4. One-Tailed vs Two-Tailed Test

● One-tailed: Testing in one direction (e.g., H₁: μ > 2)
● Two-tailed: Testing both directions (e.g., H₁: μ ≠ 2)

✅ 5. Example in Python: One Sample t-test

from scipy import stats
import numpy as np

data = np.random.normal(loc=2.1, scale=0.5, size=30)

t_stat, p_val = stats.ttest_1samp(data, popmean=2)

print("t-statistic:", t_stat)
print("p-value:", p_val)
if p_val < 0.05:
 print("Reject H₀")
else:
print("Fail to reject H₀")

✅ 6. Common Tests
Test Use Case
Z-test Known population std dev, n > 30
T-test Unknown std dev, small samples
Chi-square test Categorical variables
ANOVA Comparing >2 groups
Proportion test Proportion-based hypotheses

💬 INTERVIEW QUESTIONS (Medium to High)

🔸 Q1: What does p-value mean?
✅unlikely
Probability of observing the test results under H.₀ Low p-value ⇒
under H
₀ ⇒reject it.

🔸 Q2: How do you interpret a t-test result?

✅ If p < α, reject H₀⇒ the sample mean significantly differs from the
population mean.

🔸 Q3: When to use z-test vs t-test?

✅ Use z-test when population standard deviation is known; t-test
otherwise.

🔸 Q4: What’s the difference between one-tailed and two-tailed tests?

✅both.One-tailed tests check for deviation in one direction; two-tailed in
🔹 Day 8: T-Test Types – One-Sample, Independent,
Paired
The T-test is a fundamental statistical tool used to compare means and
check if differences are statistically significant. It's essential in A/B
testing, medical trials, and user behavior analysis.

✅ 1. When to Use T-Test?

● Population standard deviation is unknown
● Sample size is small (typically n < 30)
● Data is approximately normally distributed

✅ 2. One-Sample T-Test
Checks if the sample mean is significantly different from a known value
(usually the population mean).
from scipy.stats import ttest_1samp
import numpy as np
data = np.array([22, 21, 23, 20, 24])
t_stat, p_val = ttest_1samp(data, popmean=21)
print("T-Statistic:", t_stat)
print("P-Value:", p_val)

✅ 3. Independent Two-Sample T-Test

Checks if two independent groups have significantly different means.
Use case: Comparing conversion rates of Group A vs Group B
from scipy.stats import ttest_ind

group_a = np.random.normal(50, 5, 30)

group_b = np.random.normal(52, 5, 30)
t_stat, p_val = ttest_ind(group_a, group_b)
print("T-Statistic:", t_stat)
print("P-Value:", p_val)

✅ 4. Paired T-Test (Dependent Samples)

Used when same subjects are tested before and after a treatment.
Use case: Pre-test vs Post-test scores
from scipy.stats import ttest_rel

before = np.array([70, 68, 72, 71, 69])

after = np.array([75, 70, 73, 74, 72])
t_stat, p_val = ttest_rel(before, after)
print("T-Statistic:", t_stat)
print("P-Value:", p_val)

✅ 5. Assumptions of T-Test
● Data is continuous and normally distributed
● Observations are independent
● Equal variances (in some cases)

💬 INTERVIEW QUESTIONS (Medium to High)

🔸 Q1: What is the main difference between Independent and Paired
T-test?
✅ Independent t-test compares two different groups; paired t-test
compares the same group twice.
🔸 Q2: What are the assumptions of a T-test?
✅ Normality, independence, and equal variances (for independent t-
test).
🔸 Q3: When should you not use a T-test?
✅ When data is not normally distributed or when using
ordinal/categorical data.

🔸 Q4: Why is the T-distribution used instead of the normal

distribution?
✅ It adjusts for small sample sizes by having heavier tails, providing
more conservative results.

🔹 Day 9: Z-Test vs T-Test vs ANOVA – Know When to

Use What

Confused between Z-Test, T-Test, and ANOVA? You're not alone. These
statistical tests help determine if group differences are real or due to
chance—but each has its own use-case and assumptions.

✅ 1. Z-Test
Used when:
● Population standard deviation is known
● Sample size is large (n ≥ 30)
● Data is normally distributed
Use case: Testing population mean with large sample

from statsmodels.stats.weightstats import ztest

import numpy as np
data = np.random.normal(loc=50, scale=10, size=100)
z_stat, p_val = ztest(data, value=52)
print("Z-Statistic:", z_stat)
print("P-Value:", p_val)

✅ 2. T-Test
Used when:
● Population standard deviation is unknown
● Sample size is small (n < 30)
● Data is approximately normal

We’ve already covered:

● One-Sample T-Test
● Independent Two-Sample T-Test
● Paired T-Test

✅ 3. ANOVA (Analysis of Variance)

Used to compare means of 3 or more groups.

Use case: Comparing test scores across 3 departments

from scipy.stats import f_oneway

group1 = [88, 92, 85, 90]

group2 = [78, 79, 74, 77]
group3 = [92, 94, 96, 91]

f_stat, p_val = f_oneway(group1, group2, group3)

print("F-Statistic:", f_stat)
print("P-Value:", p_val)
✅ 4. Key Differences
Feature Z-Test T-Test ANOVA
Groups Compared 1 or 2 1 or 2 3 or more
Sample Size Large (n≥30) Small (n<30) Any
Std Dev Known? Yes No No
Output Z-statistic T-statistic F-statistic

💬 INTERVIEW QUESTIONS (Medium to High)

🔸 Q1: When should you use a Z-Test instead of a T-Test?
✅is large.
When the population standard deviation is known and sample size

🔸 Q2: Why is ANOVA better than multiple T-Tests?

✅ Reduces Type I error; multiple t-tests increase false positives.

🔸 Q3: What’s the null hypothesis in ANOVA?

✅ All group means are equal.
🔸 Q4: What do you do after a significant ANOVA result?
✅differ.
Perform post-hoc tests like Tukey’s HSD to identify which groups

🔸 Q5: Can ANOVA be used for 2 groups?

✅ Yes, but it’s equivalent to a t-test in that case.
🔹 Day 10: Correlation vs Covariance – Know the Real
Difference
Both correlation and covariance measure how two variables move
together—but they are not the same. Understanding the difference is
key for data interpretation and choosing the right statistical tools.

✅ 1. Covariance – The Raw Relationship

Covariance tells you how two variables vary together, but not how
strong the relationship is.
● Positive covariance → variables move in the same direction
● Negative covariance → variables move in opposite directions
● Magnitude is not standardized (depends on units)

import numpy as np

x = [1, 2, 3, 4, 5]
y = [2, 4, 6, 8, 10]
cov_matrix = np.cov(x, y)
print("Covariance Matrix:\n", cov_matrix)

✅ 2. Correlation – The Standardized Relationship

Correlation measures the strength and direction of the linear
relationship.
● Always between -1 and 1
● +1: Perfect positive linear relation
●
-1: Perfect negative linear relation
● 0: No linear relation

correlation = np.corrcoef(x, y)
print("Correlation Matrix:\n", correlation)
✅ 3. Key Differences Table
Feature Covariance Correlation
Range No fixed range [-1, 1]
Unit-dependent Yes No
Interpretation Direction only Direction + Strength
Standardized? No Yes
Use Cases PCA, Portfolio Analysis Feature selection, EDA

✅ 4. When to Use What?

Use Case Use This
To measure pure directional movement Covariance
To understand strength of relation Correlation
In PCA (Principal Component Analysis) Covariance
In EDA & Feature Selection Correlation

💬 INTERVIEW QUESTIONS (Easy to Medium)

🔸 Q1: Can two variables have high covariance but low correlation?
✅ Yes, if the units or scales are different, correlation may still be low.

🔸 Q2: Why do we prefer correlation in EDA?

✅ It standardizes values, making it easier to compare relationships.
Q3: What happens if correlation is 0?
🔸 There's no linear relationship, but non-linear may still exist.
✅ Q4: Is correlation always causation?

🔸 Nope! Strong correlation ≠ cause-effect relationship.

✅
🔹 Day 11: Chi-Square Test – For Categorical Data Only
The Chi-Square Test (χ²) is a powerful statistical method used to
determine if there’s a significant association between categorical
variables. It’s your go-to test when analyzing survey results,
demographics, marketing funnels, and more!

✅ 1. What is the Chi-Square Test?

● Tests independence between two categorical variables
● Compares observed frequencies to expected frequencies

✅ 2. Types of Chi-Square Tests

Test Type Use Case
Chi-Square of Independence Are two categorical variables related?
Chi-Square Goodness of Fit Does a sample match a population?

✅ 4. Using Chi-Square in Python (SciPy)

import pandas as pd
from scipy.stats import chi2_contingency
# Sample contingency table
data = [[30, 10],
[20, 40]]

chi2, p, dof, expected = chi2_contingency(data)

print("Chi-Square Value:", chi2)

print("P-Value:", p)
print("Degrees of Freedom:", dof)
print("Expected Frequencies:\n", expected)
✅ 5. Interpretation

● p < 0.05 → Reject null hypothesis → Significant association

● p ≥ 0.05→ Fail to reject → No significant association

📌 Applications
● Is gender related to product preference?
● Are age groups independent of newsletter signup?
● Do regions and voting patterns relate?

💬 INTERVIEW QUESTIONS (Medium Level)

🔸 Q1: Can Chi-Square be used for numerical data?

✅ No. It’s strictly for categorical data.

🔸 Q2: What are the assumptions of Chi-Square?

Observations should be
✅ independent, and expected frequency
≥ 5 in each cell (preferably).
should be
🔸 Q3: What’s the role of degrees of freedom?
✅ It determines the distribution curve and affects p-value calculation.
🔸 Q4: What if the expected frequency is < 5 in many cells?
✅ Use Fisher’s Exact Test instead.

🔹 Day 12: Central Limit Theorem (CLT) – The Heart of

Statistics ❤️

The Central Limit Theorem (CLT) is one of the most fundamental ideas
in statistics and data science. It explains why normal distribution
appears so often, even when the data itself isn’t normal!

✅ 1. What is the Central Limit Theorem?

No matter the original distribution of the data, the sampling distribution
of the sample mean will approach a normal distribution as the sample
size becomes large (typically n ≥ 30).

✅ 2. Why is CLT so Important?

● It allows parametric tests (like Z-tests and t-tests) even on non-
normal data
● Powers confidence intervals and hypothesis testing
● Makes large-sample inference possible!

✅ 3. CLT in Action: Python Example

import numpy as np
import matplotlib.pyplot as plt
# Generate skewed data (exponential)
data = np.random.exponential(scale=2, size=10000)
sample_means = []

# Take 1000 samples of size 50 each

for _ in range(1000):
sample = np.random.choice(data, size=50)
sample_means.append(np.mean(sample))

# Plot the distribution of sample means

plt.hist(sample_means, bins=30, color='skyblue', edgecolor='black')
plt.title("Sampling Distribution Approaches Normality")
plt.xlabel("Sample Mean")
plt.ylabel("Frequency")
plt.show()

✅ 4. Key Terms
Term Meaning
Population Entire group
Sample Subset of the population
Sampling Distribution Distribution of sample statistics

✅ 5. CLT Assumptions
● Samples are random and independent
● Sample size n ≥ 30 is usually sufficient
● Population must have finite variance

📌 Applications of CLT
● A/B testing
● Quality control in manufacturing
● Estimating population mean from sample
● Predictive analytics in ML workflows
💬 INTERVIEW QUESTIONS (Medium-High)
🔸 Q1: Why is the Central Limit Theorem important for hypothesis
testing?
✅ Because it allows us to use the normal distribution even when
data isn't normal!

🔸 Q2: What’s the difference between population distribution and

sampling distribution?
✅ Population: Original data distribution.
✅ Sampling: Distribution of sample means.
🔸 It primarily applies to sample means.
Q3: Does CLT apply to the median or standard deviation?

✅
🔸 Q4: What if the population is already normal?
✅sample
Then the sampling distribution of the mean is also normal for any
size.

🔹 Day 13: t-Tests – One Sample, Two Sample, Paired t-

Test
The t-Test is your go-to method when comparing means—whether it's
comparing a group to a benchmark, or comparing two
related/unrelated groups.

✅ 1. What is a t-Test?
A t-Test determines whether the difference in means is statistically
significant when the sample size is small and the population standard
deviation is unknown.
✅ 2. Types of t-Tests
Test Type
Use Case
One Sample t-Test
Compare one group mean to a known
Two Sample value
(Independent) t-Test
Paired Sample t-Test Compare two independent group means
(e.g., A/B testing)
Compare two related groups (e.g.,
before vs. after)

✅ 3. Assumptions of t-Test
● Data is normally distributed (CLT helps here!)
● Observations are independent
● Variances are equal (for two-sample t-test)
● Data is measured on interval/ratio scale

✅ 4. Python Examples
📌 One-Sample t-Test
from scipy import stats
import numpy as np
sample = [23, 25, 21, 22, 24, 26]
t_stat, p_val = stats.ttest_1samp(sample, popmean=20)
print("t-statistic:", t_stat, "p-value:", p_val)

📌 Two-Sample (Independent) t-Test

group1 = [30, 32, 29, 35, 28]
group2 = [25, 26, 27, 30, 22]
t_stat, p_val = stats.ttest_ind(group1, group2)
print("t-statistic:", t_stat, "p-value:", p_val)
📌 Paired t-Test
before = [80, 85, 88, 90]
after = [82, 87, 89, 91]
t_stat, p_val = stats.ttest_rel(before, after)
print("t-statistic:", t_stat, "p-value:", p_val)

✅ 5. Interpreting p-Value
p-value Interpretation
p < 0.05 Reject the null hypothesis (significant)
p ≥ 0.05 Fail to reject the null (not significant)

📌 Real-World Applications
● A/B Testing (website layout A vs. layout B)
● Medical Studies (treatment vs. control)
● Before/After Improvements (new teaching method vs. old)

💬 INTERVIEW QUESTIONS (Medium to High)

🔸 Q1: What is the difference between independent and paired t-
tests?
✅ Independent: different groups; Paired: same group at two times.
🔸 When assumptions (normality, independence, equal variance) are
Q2: When can you not use a t-test?

✅violated.
🔸 Q3: What’s the null hypothesis in a t-test?
✅ That the means are equal (no significant difference).
🔸 Q4: Why use t-distribution instead of normal?
✅ t-distribution accounts for extra uncertainty in small samples.

🔹 Day 14: ANOVA – Analysis of Variance

ANOVA (Analysis of Variance) is your go-to test when you want to
compare the means of 3 or more groups at once instead of doing
multiple t-tests.

✅ 1. What is ANOVA?
ANOVA helps us determine whether the mean differences across
multiple groups are statistically significant.
It avoids the risk of Type I error that comes from performing multiple t-
tests.

✅ 2. Types of ANOVA
Type Use Case
One-Way 1 categorical IV (grouping) → 1 continuous DV
ANOVA (outcome)
Two-Way
2 categorical IVs → 1 continuous DV
ANOVA
Same subjects measured multiple times (like
Repeated
paired t-test)
Measures

✅ 3. Assumptions of ANOVA
● Groups are independent
● Normal distribution of the outcome variable
● Equal variances across groups (homogeneity of variances)
● Outcome variable is continuous
✅ 4. Python Example: One-Way ANOVA
from scipy.stats import f_oneway

group_A = [25, 30, 28, 35, 33]

group_B = [22, 29, 27, 24, 26]
group_C = [35, 40, 38, 37, 42]

f_stat, p_val = f_oneway(group_A, group_B, group_C)

print("F-statistic:", f_stat, "p-value:", p_val)

✅ 5. Interpreting Results
● F-statistic: Ratio of between-group variance to within-group
variance
● p-value:

o p < 0.05 → At least one group mean is significantly different

o p ≥ 0.05 → No significant difference between group means

📌 Post-hoc Tests (If ANOVA is Significant)

Use Tukey’s HSD or Bonferroni test to find which pairs of groups differ
significantly.

📈 Real-World Applications
● Comparing test scores across 3 different teaching methods
● Measuring user engagement across 3 website layouts
● Comparing sales performance in different regions
💬 INTERVIEW QUESTIONS (Medium to High)
🔸 Q1: Why not just use multiple t-tests instead of ANOVA?
✅ Multiple t-tests increase the chance of Type I error.
🔸 Q2: What does a high F-statistic mean?
Greater variation
✅groups. between group means compared to within

🔸 Q3: When to use ANOVA over t-test?

✅ When comparing 3 or more group means.

🔸 Q4: What should you do after a significant ANOVA result?

✅ Perform a post-hoc test to identify which groups differ.

🔹 Day 15: Chi-Square Test – Categorical Data Analysis

The Chi-Square (χ²) Test helps us analyze relationships between

categorical variables. It’s widely used in surveys, marketing, healthcare,
and more!

✅ 1. What is Chi-Square Test?

It checks whether two categorical variables are independent or
associated.
📌 Example:
Does gender influence product preference?

✅ 2. Types of Chi-Square Tests

Type Use Case
Goodness of Fit
One categorical variable vs expected
Test of
distribution
Independence
Two categorical variables (e.g., Gender vs
Choice)

✅ 3. Assumptions of Chi-Square
● Data should be frequencies (counts), not percentages
● Categories must be mutually exclusive
● Expected frequency in each cell should be ≥ 5 for validity
● Observations are independent

✅ 4. Python Example: Test of Independence

import pandas as pd
from scipy.stats import chi2_contingency
# Contingency table
data = pd.DataFrame({
'Product A': [30, 10],
'Product B': [20, 40]
}, index=['Male', 'Female'])

chi2, p, dof, expected = chi2_contingency(data)

print("Chi2 Statistic:", chi2)
print("p-value:", p)
print("Degrees of Freedom:", dof)
print("Expected Frequencies:\n", expected)

✅ 5. Interpreting Results
● Chi² value: Larger = more difference between observed and
expected
● p-value:
 o p < 0.05 → Variables are likely associated
o p ≥ 0.05 → Variables are likely independent

📈 Real-World Applications
● Is gender associated with product preference?
● Does education level impact voting behavior?
● Do different ads get clicked by different age groups?

💬 INTERVIEW QUESTIONS (Medium to High)

🔸 Q1: Can Chi-Square test be used on numerical data?
✅ No, it's only for categorical data.
🔸 Q2: What’s the difference between Chi-Square Goodness-of-Fit
and Test of Independence?
✅ Goodness-of-Fit → One variable; Independence → Two variables.
🔸 Q3: Why should expected frequency be ≥ 5?

✅ To ensure accuracy of the Chi-Square approximation.

🔸 Q4: What does a low p-value mean in Chi-Square?

✅ There's a statistical association between the variables.

👏 Congratulations! You've completed the 15-Day Statistics Crash

Course!
📚 You now have hands-on knowledge of:
● Descriptive & Inferential Stats
● Central Tendency & Dispersion
● Distributions
● Hypothesis Testing
● ANOVA & Chi-Square
20 Medium-level Statistics Interview Questions
📊 1. What is the Central Limit Theorem and why is it important?
The Central Limit Theorem states that the sampling distribution of the
sample mean approaches a normal distribution as the sample size
becomes large, regardless of the population's distribution. This is crucial
because it allows for the use of inferential statistics, such as confidence
intervals and hypothesis tests, even when the population distribution is
unknown.

📈 2. Explain the difference between Type I and Type II errors.

● Type I Error (False Positive): Rejecting the null hypothesis when it
is actually true.
● Type II Error (False Negative): Failing to reject the null hypothesis
when it is actually false.
Understanding these errors is vital for evaluating the reliability of
statistical tests.

🧪 3. How do you interpret a p-value?

A p-value represents the probability of obtaining a test statistic at least
as extreme as the one observed, assuming the null hypothesis is true. A
smaller p-value indicates stronger evidence against the null hypothesis.
Typically, a p-value less than 0.05 is considered statistically significant.
Learn R, Python & Data Science Online+1DataLemur+1

📊 4. What is the difference between a t-test and a z-test?

● Z-test: Used when the population variance is known and the
sample size is large (n > 30).
● T-test: Used when the population variance is unknown and/or the
sample size is small (n ≤ 30).
Both tests assess whether there is a significant difference between
sample means.

📉 5. Describe the assumptions of linear regression.

1. Linearity: The relationship between independent and dependent
variables is linear.
2. Independence: Observations are independent of each other.
3. Homoscedasticity: Constant variance of errors.
4. Normality: The residuals (errors) are normally distributed.
5. No multicollinearity: Independent variables are not highly
correlated with each other.

📈 6. What is multicollinearity and how can it be detected?

Multicollinearity occurs when independent variables in a regression
model are highly correlated, leading to unreliable coefficient estimates.
It can be detected using:
● Variance Inflation Factor (VIF): Values greater than 5 or 10 indicate
multicollinearity.
● Correlation Matrix: High correlation coefficients between variables
suggest multicollinearity.

🧮 7. Explain the difference between correlation and causation.

● Correlation: Measures the strength and direction of a linear
relationship between two variables.
● Causation: Implies that one variable directly affects another.

Correlation does not imply causation; two variables may be correlated

due to coincidence or a third variable.

📊 8. What is the purpose of an ANOVA test?

ANOVA (Analysis of Variance) tests whether there are statistically
significant differences between the means of three or more
independent groups. It analyzes the variance within groups and
between groups to determine if at least one group mean is different.

📈 9. How do you interpret R-squared and adjusted R-squared in

regression analysis?

● R-squared: Represents the proportion of variance in the

dependent variable explained by the independent variables.
● Adjusted R-squared: Adjusts R-squared for the number of
predictors in the model, providing a more accurate measure when
comparing models with different numbers of predictors.

🧪 10. What is a p-value and how is it used in hypothesis testing?

A p-value is the probability of obtaining test results at least as extreme
as the observed results, assuming the null hypothesis is true. In
hypothesis testing, if the p-value is less than the chosen significance
level (e.g., 0.05), the null hypothesis is rejected.Learn R, Python & Data
Science Online

📊 11. Describe the difference between a one-tailed and two-tailed

test.

● One-tailed test: Tests for the possibility of the relationship in one

direction.
● Two-tailed test: Tests for the possibility of the relationship in both
directions.Data Interview
The choice depends on the research hypothesis.

📈 12. What is the difference between parametric and non-parametric

tests?

● Parametric tests: Assume underlying statistical distributions (e.g.,

normal distribution).
 ● Non-parametric tests: Do not assume specific distributions and
are used when data doesn't meet parametric test assumptions.

🧮 13. Explain the concept of statistical power.

Statistical power is the probability that a test correctly rejects a false
null hypothesis (i.e., detects an effect when there is one). Higher power
reduces the risk of Type II errors.

📊 14. What is the difference between a population and a sample?

● Population: The entire group being studied.
● Sample: A subset of the population used to make inferences about
the whole.Data Interview

📈 15. Describe the difference between descriptive and inferential

statistics.

● Descriptive statistics: Summarize and describe the features of a

dataset.
● Inferential statistics: Make predictions or inferences about a
population based on a sample.

🧪 16. What is a confidence interval?

A confidence interval is a range of values, derived from the sample data,
that is likely to contain the value of an unknown population parameter.
For example, a 95% confidence interval suggests that there is a 95%
chance the interval contains the true parameter.
📊 17. What is stratified sampling, and when is it used?
Stratified sampling involves dividing the population into distinct
subgroups (strata) based on shared characteristics and then sampling
from each stratum proportionally. This method ensures representation
from all subgroups, leading to more precise estimates, especially when
there are significant differences between strata. DataLemur+1Learn R,
Python & Data Science Online+1

📈 18. Explain the difference between a population parameter and a

sample statistic.

● Population Parameter: A value that describes a characteristic of

the entire population (e.g., population mean μ).
● Sample Statistic: A value that describes a characteristic of a
sample drawn from the population (e.g., sample mean x̄̄ ).

Sample statistics are used to estimate population parameters.Learn R,

Python & Data Science Online+1GitHub+1

🧮 19. What is the law of large numbers?

The law of large numbers states that as the size of a sample increases,
the sample mean will get closer to the population mean. This principle
underpins many statistical practices, ensuring that larger samples
provide more accurate estimates.

📊 20. How do you determine if a dataset is normally distributed?

To assess normality:
● Visual Methods:
o Histogram: Should resemble a bell-shaped curve.
o Q-Q Plot: Data points should lie approximately along the
reference line.Learn R, Python & Data Science Online
 ● Statistical Tests:
o Shapiro-Wilk Test: Tests the null hypothesis that the data is
normally distributed.
o Kolmogorov-Smirnov Test: Compares the sample
distribution with a normal distribution.

A combination of visual and statistical methods provides a

comprehensive assessment of normality.
20 Hard-level Statistics Interview Questions
1. What is the difference between Maximum Likelihood Estimation
(MLE) and Bayesian Estimation?
MLE seeks parameter values that maximize the likelihood function,
relying solely on observed data. Bayesian Estimation incorporates prior
beliefs and updates them with observed data to form a posterior
distribution.

2. Explain the concept of the Bias-Variance Tradeoff.

The Bias-Variance Tradeoff describes the balance between:Data

Interview
● Bias: Error from erroneous assumptions in the learning algorithm.
● Variance: Error from sensitivity to small fluctuations in the training
set.GitHub+3Learn R, Python & Data Science Online+3Data
Interview+3
High bias can cause underfitting, while high variance can cause
overfitting. The goal is to find a model with optimal complexity to
minimize total error.

3. What is the purpose of regularization in regression models?

Regularization adds a penalty term to the loss function to prevent
overfitting by discouraging complex models:
● Lasso Regression (L1): Encourages sparsity, potentially reducing
some coefficients to zero.
 ● Ridge Regression (L2): Shrinks coefficients towards zero but
doesn't eliminate them.

4. How do you interpret the Area Under the ROC Curve (AUC-ROC)?
AUC-ROC measures a classifier's ability to distinguish between classes:
● AUC = 1: Perfect classification.
● AUC = 0.5: No discriminative ability (equivalent to random
guessing).
Higher AUC indicates better model performance.
5. What is the Central Limit Theorem and why is it important?
The Central Limit Theorem states that the sampling distribution of the
sample mean approaches a normal distribution as the sample size
increases, regardless of the population's distribution. This is crucial for
making inferences about population parameters.

6. Explain the concept of multicollinearity and how to detect it.

Multicollinearity occurs when independent variables in a regression
model are highly correlated, leading to unreliable coefficient estimates.
It can be detected using:
● Variance Inflation Factor (VIF): Values above 5 or 10 indicate high
multicollinearity.
● Correlation Matrix: High correlation coefficients between variables
suggest multicollinearity.

7. What is the difference between parametric and non-parametric

tests?
● Parametric Tests: Assume underlying statistical distributions (e.g.,
t-test assumes normality).
● Non-Parametric Tests: Make fewer assumptions about data
distribution (e.g., Mann-Whitney U test).
Non-parametric tests are useful when data doesn't meet parametric
test assumptions.
8. Describe the concept of overfitting and how to prevent it.
Overfitting occurs when a model learns noise in the training data,
performing well on training data but poorly on unseen data. Prevention
techniques include:
● Cross-validation: Ensures model generalization.
● Regularization: Adds penalty terms to discourage complexity.
● Pruning: Reduces complexity in decision trees.
● Early Stopping: Halts training when performance on validation set
deteriorates.

9. What is the difference between Type I and Type II errors?

● Type I Error (False Positive): Rejecting a true null hypothesis.
● Type II Error (False Negative): Failing to reject a false null
hypothesis.DataLemur+7LinkedIn+7Learn R, Python & Data Science
Online+7
Balancing these errors is crucial in hypothesis testing.

10. How do you handle missing data in a dataset?

Strategies include:
● Deletion: Removing records with missing values (listwise or
pairwise).
● Imputation: Filling in missing values using methods like mean,
median, mode, or predictive models.
● Model-Based Methods: Using algorithms that handle missing data
inherently.
The choice depends on the nature and extent of missingness.

11. Explain the concept of statistical power.

Statistical power is the probability of correctly rejecting a false null
hypothesis (i.e., detecting an effect when it exists). Higher power
reduces the risk of Type II errors. Factors influencing power include
sample size, effect size, significance level, and variability.
12. What is the purpose of cross-validation in model evaluation?
Cross-validation assesses a model's ability to generalize to unseen data
by partitioning the data into training and validation sets multiple times.
Common methods include k-fold and leave-one-out cross-validation.
It helps in selecting models and tuning hyperparameters.

13. Describe the differences between bagging and boosting.

● Bagging (Bootstrap Aggregating): Builds multiple independent
models on random subsets of data and aggregates their
predictions (e.g., Random Forest).
● Boosting: Builds models sequentially, each trying to correct the
errors of the previous one (e.g., AdaBoost, Gradient Boosting).
Bagging reduces variance, while boosting reduces bias.

14. What is Principal Component Analysis (PCA) and when is it used?

PCA is a dimensionality reduction technique that transforms correlated
variables into a set of uncorrelated components, ordered by the amount
of variance they capture. It's used to simplify datasets, reduce noise,
and prevent overfitting.