Complete Statistics Notes with Python

1. What is Statistics?
Statistics is the subject used to collect, organize, analyze, interpret, and present data.

In Data Science, statistics helps us answer questions like:

• What is the average sales value?

• Are two groups different from each other?
• Is there a relationship between two columns?
• Can we trust a sample result?
• Is a machine learning model performing well?

Example:

import pandas as pd

sales = [100, 150, 200, 250, 300]

df = [Link]({"sales": sales})
print(df)

2. Types of Statistics
Statistics is mainly divided into two types:

2.1 Descriptive Statistics

Descriptive statistics describes the data that we already have.

Example questions:

• What is the mean?

• What is the median?
• What is the highest value?
• How spread out is the data?

2.2 Inferential Statistics

Inferential statistics uses sample data to make conclusions about a larger population.

1
Example questions:

• Is the average salary of data analysts greater than 40,000?

• Are male and female customers spending differently?
• Is a new website design improving conversion rate?

3. Population and Sample

Population
The full group you want to study.

Example:

All Netflix users in India.

Sample
A smaller part selected from the population.

Example:

1,000 Netflix users selected from India.

Why sample?

Because studying the entire population is usually expensive, slow, or impossible.

import pandas as pd

population = [Link]({
"user_id": range(1, 101),
"watch_hours": range(10, 110)
})

sample = [Link](n=10, random_state=42)

print(sample)

2
4. Types of Data
4.1 Numerical Data
Data represented using numbers.

Examples:

• Age
• Salary
• Price
• Rating
• Sales

Numerical data has two types:

Discrete Data

Countable numbers.

Examples:

• Number of students
• Number of apps
• Number of movies

Continuous Data

Measurable values that can contain decimals.

Examples:

• Height
• Weight
• Temperature
• Price

4.2 Categorical Data

Data represented using categories.

Examples:

• Gender
• City
• Movie genre

3
 • App category
• Payment type

Categorical data has two types:

Nominal Data

Categories without order.

Examples:

• Gender
• City
• Genre

Ordinal Data

Categories with order.

Examples:

• Low, Medium, High

• Poor, Average, Good, Excellent

5. Creating a Simple Dataset

import pandas as pd

_data = {
"student": ["A", "B", "C", "D", "E", "F", "G", "H"],
"gender": ["Male", "Female", "Male", "Female", "Male", "Female", "Male",
"Female"],
"study_hours": [2, 4, 3, 5, 6, 7, 4, 8],
"marks": [45, 60, 55, 70, 75, 85, 65, 90],
"passed": ["No", "Yes", "Yes", "Yes", "Yes", "Yes", "Yes", "Yes"]
}

df = [Link](_data)
print(df)

4
6. Descriptive Statistics
6.1 Mean
Mean means average.

Formula:

Mean = Sum of all values / Number of values

mean_marks = df["marks"].mean()
print(mean_marks)

Use mean when:

• Data is numerical
• Data does not have many extreme outliers

Example:

Average marks of students.

6.2 Median
Median is the middle value after sorting the data.

median_marks = df["marks"].median()
print(median_marks)

Use median when:

• Data has outliers

• You want a more stable center value

Example:

Median salary is usually better than mean salary because some people may have extremely high salaries.

6.3 Mode
Mode is the most repeated value.

5
 mode_gender = df["gender"].mode()
print(mode_gender)

Use mode when:

• Data is categorical
• You want the most common category

Example:

Most common app category.

6.4 Minimum and Maximum

print(df["marks"].min())
print(df["marks"].max())

6.5 Range
Range shows the difference between maximum and minimum values.

marks_range = df["marks"].max() - df["marks"].min()

print(marks_range)

Range is simple but sensitive to outliers.

7. Variance and Standard Deviation

7.1 Variance
Variance tells how much values are spread from the mean.

High variance means values are far from the mean.

Low variance means values are close to the mean.

6
 variance_marks = df["marks"].var()
print(variance_marks)

7.2 Standard Deviation

Standard deviation is the square root of variance.

It is easier to understand because it is in the same unit as the original data.

std_marks = df["marks"].std()
print(std_marks)

Interpretation:

If average marks are 68 and standard deviation is 15, then marks usually vary around 15 marks from the
average.

8. Percentiles and Quartiles

Percentile
A percentile tells the position of a value in the data.

Example:

If a student is at the 90th percentile, it means the student scored better than 90% of students.

print(df["marks"].quantile(0.25))
print(df["marks"].quantile(0.50))
print(df["marks"].quantile(0.75))

Quartiles
Quartiles divide data into 4 parts.

• Q1 = 25th percentile
• Q2 = 50th percentile = median
• Q3 = 75th percentile

7
9. Interquartile Range IQR
IQR measures the spread of the middle 50% data.

Formula:

IQR = Q3 - Q1

Q1 = df["marks"].quantile(0.25)
Q3 = df["marks"].quantile(0.75)
IQR = Q3 - Q1
print(IQR)

IQR is useful for detecting outliers.

10. Outlier Detection using IQR

Outliers are extreme values.

Formula:

Lower Limit = Q1 - 1.5 * IQR

Upper Limit = Q3 + 1.5 * IQR

Q1 = df["marks"].quantile(0.25)
Q3 = df["marks"].quantile(0.75)
IQR = Q3 - Q1

lower_limit = Q1 - 1.5 * IQR

upper_limit = Q3 + 1.5 * IQR

outliers = df[(df["marks"] < lower_limit) | (df["marks"] > upper_limit)]

print(outliers)

8
11. Summary Statistics in Pandas

print([Link]())

For categorical columns:

print([Link](include="object"))

For all columns:

print([Link](include="all"))

12. Frequency Count

Frequency means how many times each value appears.

print(df["gender"].value_counts())
print(df["passed"].value_counts())

Percentage frequency:

print(df["gender"].value_counts(normalize=True) * 100)

13. GroupBy Statistics

GroupBy is used to calculate statistics for each category.

Example:

Average marks by gender.

print([Link]("gender")["marks"].mean())

Multiple statistics:

9
 print([Link]("gender")["marks"].agg(["mean", "median", "min", "max",
"std"]))

14. Covariance
Covariance tells whether two numerical variables move together.

If covariance is positive:

• Both variables increase together.

If covariance is negative:

• One variable increases and the other decreases.

If covariance is near zero:

• There may be little or no linear relationship.

cov_value = df["study_hours"].cov(df["marks"])
print(cov_value)

Problem with covariance:

Its value depends on the units of data, so it is difficult to compare.

15. Correlation
Correlation measures the strength and direction of relationship between two numerical variables.

Correlation value is always between -1 and +1.

• +1 means perfect positive relationship

• -1 means perfect negative relationship
• 0 means no linear relationship

corr_value = df["study_hours"].corr(df["marks"])
print(corr_value)

Correlation matrix:

10
 print(df[["study_hours", "marks"]].corr())

Use correlation when:

• You want to understand relationship strength

• You want to compare relationships between different columns

16. Visualizing Statistics

import [Link] as plt

[Link](df["marks"])
[Link]("Marks")
[Link]("Frequency")
[Link]("Distribution of Marks")
[Link]()

Boxplot:

[Link](df["marks"])
[Link]("Marks")
[Link]("Boxplot of Marks")
[Link]()

Scatter plot:

[Link](df["study_hours"], df["marks"])
[Link]("Study Hours")
[Link]("Marks")
[Link]("Study Hours vs Marks")
[Link]()

17. Probability Basics

Probability measures the chance of an event happening.

Formula:

11
Probability = Favorable Outcomes / Total Outcomes

Example:

Probability of getting heads in a coin toss = 1 / 2 = 0.5

prob_head = 1 / 2
print(prob_head)

18. Random Variables

A random variable is a variable whose value depends on the outcome of a random event.

Example:

If we roll a dice, X can be 1, 2, 3, 4, 5, or 6.

import random

x = [Link](1, 6)
print(x)

19. Probability Distributions

A probability distribution tells how probabilities are distributed over possible values.

There are two main types:

19.1 Discrete Probability Distribution

Used for countable outcomes.

Examples:

• Dice roll
• Number of customers
• Number of clicks

12
19.2 Continuous Probability Distribution
Used for measurable values.

Examples:

• Height
• Weight
• Salary
• Time

20. Uniform Distribution

In uniform distribution, all outcomes have equal probability.

Example:

Rolling a fair dice.

import numpy as np

uniform_data = [Link](low=0, high=10, size=1000)

[Link](uniform_data, bins=30)
[Link]("Uniform Distribution")
[Link]()

21. Normal Distribution

Normal distribution is a bell-shaped distribution.

Many real-world values roughly follow normal distribution.

Examples:

• Height
• Exam marks
• Measurement errors

13
 normal_data = [Link](loc=50, scale=10, size=1000)

[Link](normal_data, bins=30)
[Link]("Normal Distribution")
[Link]()

Here:

• loc = mean
• scale = standard deviation

22. Standard Normal Distribution

Standard normal distribution has:

• Mean = 0
• Standard deviation = 1

standard_normal = [Link](loc=0, scale=1, size=1000)

[Link](standard_normal, bins=30)
[Link]("Standard Normal Distribution")
[Link]()

23. Z-Score
Z-score tells how far a value is from the mean in terms of standard deviation.

Formula:

Z = (X - Mean) / Standard Deviation

x = 85
mean = df["marks"].mean()
std = df["marks"].std()

z_score = (x - mean) / std

print(z_score)

14
Interpretation:

• Z-score = 0 means value is equal to mean

• Positive z-score means value is above mean
• Negative z-score means value is below mean
• Z-score above 3 or below -3 can be considered extreme

24. Central Limit Theorem

Central Limit Theorem says:

If we take many samples from a population and calculate their means, the distribution of sample means will
become approximately normal, even if the original population is not normal.

This is very important because many hypothesis tests depend on normality.

population = [Link](scale=2, size=10000)

sample_means = []

for i in range(1000):
sample = [Link](population, size=30)
sample_means.append([Link]())

[Link](sample_means, bins=30)
[Link]("Central Limit Theorem")
[Link]()

25. Sampling Techniques

25.1 Random Sampling
Every row has equal chance of selection.

sample_df = [Link](n=4, random_state=42)

print(sample_df)

15
25.2 Stratified Sampling
Data is sampled while maintaining category proportions.

Useful for imbalanced data.

from sklearn.model_selection import train_test_split

train_df, test_df = train_test_split(

df,
test_size=0.25,
stratify=df["passed"],
random_state=42
)

print(train_df["passed"].value_counts(normalize=True))
print(test_df["passed"].value_counts(normalize=True))

26. Hypothesis Testing

Hypothesis testing is used to check whether a claim about data is statistically supported.

Important Terms

Null Hypothesis H0

The default assumption.

Usually says there is no effect, no difference, or no relationship.

Alternative Hypothesis H1

The claim we want to test.

Usually says there is an effect, difference, or relationship.

Alpha

Alpha is the significance level.

Common value:

16
 alpha = 0.05

This means we accept a 5% chance of making a wrong rejection of the null hypothesis.

P-value

P-value tells how likely the observed result is if the null hypothesis is true.

Decision rule:

If p-value < alpha: Reject H0

If p-value >= alpha: Fail to reject H0

27. One-Tailed and Two-Tailed Tests

Left-Tailed Test
Used when the alternative hypothesis says the value is less than something.

Example:

Average app size is less than 25 MB.

Right-Tailed Test
Used when the alternative hypothesis says the value is greater than something.

Example:

Average rating is greater than 4.0.

Two-Tailed Test
Used when the alternative hypothesis says the value is different from something.

Example:

Average salary is different from 40,000.

17
28. Z-Test
Z-test is used when:

• Sample size is large, usually n >= 30

• Population standard deviation is known, or sample is large enough
• Data is numerical

Install statsmodels if needed:

pip install statsmodels

from [Link] import ztest

marks = df["marks"]

z_stat, p_value = ztest(

marks,
value=60,
alternative="larger"
)

alpha = 0.05

print("Z Statistic:", z_stat)

print("P-value:", p_value)

if p_value < alpha:

print("Reject Null Hypothesis")
else:
print("Fail to Reject Null Hypothesis")

Business question:

Is the average student mark significantly greater than 60?

29. T-Test
T-test is used when:

• Sample size is small

• Population standard deviation is unknown

18
 • Data is numerical

29.1 One Sample T-Test

Used to compare sample mean with a fixed value.

from [Link] import ttest_1samp

marks = df["marks"]

t_stat, p_value = ttest_1samp(marks, popmean=60)

print("T Statistic:", t_stat)

print("P-value:", p_value)

if p_value < alpha:

print("Reject Null Hypothesis")
else:
print("Fail to Reject Null Hypothesis")

Business question:

Is the average mark different from 60?

29.2 Independent Two Sample T-Test

Used to compare means of two independent groups.

Example:

Do male and female students have different average marks?

from [Link] import ttest_ind

male_marks = df[df["gender"] == "Male"]["marks"]

female_marks = df[df["gender"] == "Female"]["marks"]

t_stat, p_value = ttest_ind(male_marks, female_marks, equal_var=False)

print("T Statistic:", t_stat)

print("P-value:", p_value)

if p_value < alpha:

19
 print("Reject Null Hypothesis")
else:
print("Fail to Reject Null Hypothesis")

Use equal_var=False for Welch's t-test, which is safer when group variances may be different.

29.3 Paired T-Test

Used when the same subjects are measured twice.

Example:

Marks before training and after training.

from [Link] import ttest_rel

before = [50, 55, 60, 65, 70]

after = [60, 58, 65, 70, 78]

t_stat, p_value = ttest_rel(before, after)

print("T Statistic:", t_stat)

print("P-value:", p_value)

if p_value < alpha:

print("Reject Null Hypothesis")
else:
print("Fail to Reject Null Hypothesis")

30. ANOVA
ANOVA means Analysis of Variance.

It is used to compare the means of more than two groups.

Example:

Do students from different teaching methods have different average marks?

20
 anova_df = [Link]({
"method": ["Online", "Online", "Online", "Offline", "Offline", "Offline",
"Hybrid", "Hybrid", "Hybrid"],
"marks": [70, 75, 72, 65, 68, 66, 80, 82, 78]
})

print(anova_df)

from [Link] import f_oneway

online = anova_df[anova_df["method"] == "Online"]["marks"]

offline = anova_df[anova_df["method"] == "Offline"]["marks"]
hybrid = anova_df[anova_df["method"] == "Hybrid"]["marks"]

f_stat, p_value = f_oneway(online, offline, hybrid)

print("F Statistic:", f_stat)

print("P-value:", p_value)

if p_value < alpha:

print("Reject Null Hypothesis")
else:
print("Fail to Reject Null Hypothesis")

Interpretation:

If p-value < 0.05, at least one group mean is significantly different.

Important:

ANOVA tells that a difference exists, but it does not directly tell which groups are different.

31. Chi-Square Test

Chi-square test is used for categorical variables.

It checks whether two categorical columns are related.

Example:

Is gender related to pass/fail result?

21
 from [Link] import chi2_contingency

contingency_table = [Link](df["gender"], df["passed"])

print(contingency_table)

chi2_stat, p_value, dof, expected = chi2_contingency(contingency_table)

print("Chi-square Statistic:", chi2_stat)

print("P-value:", p_value)
print("Degrees of Freedom:", dof)
print("Expected Values:")
print(expected)

if p_value < alpha:

print("Reject Null Hypothesis")
else:
print("Fail to Reject Null Hypothesis")

Interpretation:

If p-value < 0.05, the two categorical variables are significantly related.

32. Correlation Test

Correlation test checks whether two numerical columns are significantly related.

from [Link] import pearsonr

corr, p_value = pearsonr(df["study_hours"], df["marks"])

print("Correlation:", corr)
print("P-value:", p_value)

if p_value < alpha:

print("Reject Null Hypothesis")
else:
print("Fail to Reject Null Hypothesis")

Business question:

Is there a significant relationship between study hours and marks?

22
33. Pearson, Spearman, and Kendall Correlation
Pearson Correlation
Used when:

• Both variables are numerical

• Relationship is linear
• Data is roughly normally distributed

print(df["study_hours"].corr(df["marks"], method="pearson"))

Spearman Correlation
Used when:

• Relationship is monotonic
• Data may not be normally distributed
• Data may be ordinal

print(df["study_hours"].corr(df["marks"], method="spearman"))

Kendall Correlation
Used when:

• Dataset is small
• Data is ordinal

print(df["study_hours"].corr(df["marks"], method="kendall"))

34. Confidence Interval

A confidence interval gives a range where the true population value is likely to exist.

Example:

A 95% confidence interval for average marks may be:

23
 60 to 75

This means we are reasonably confident that the true average mark lies between 60 and 75.

import [Link] as stats

import numpy as np

marks = df["marks"]

confidence = 0.95
n = len(marks)
mean = [Link](marks)
std_error = [Link](marks)

ci = [Link](
confidence=confidence,
df=n-1,
loc=mean,
scale=std_error
)

print(ci)

35. Skewness
Skewness tells whether data is symmetric or tilted.

print(df["marks"].skew())

Interpretation:

• Skewness around 0 means data is nearly symmetric

• Positive skew means long tail on right side
• Negative skew means long tail on left side

36. Kurtosis
Kurtosis tells about the heaviness of tails in a distribution.

24
 print(df["marks"].kurt())

High kurtosis means more extreme values.

Low kurtosis means fewer extreme values.

37. Normality Test

Normality tests check whether data follows a normal distribution.

Shapiro-Wilk Test

from [Link] import shapiro

stat, p_value = shapiro(df["marks"])

print("Statistic:", stat)
print("P-value:", p_value)

if p_value < alpha:

print("Data is not normally distributed")
else:
print("Data looks normally distributed")

38. Standardization
Standardization converts data to mean 0 and standard deviation 1.

Formula:

Z = (X - Mean) / Standard Deviation

Use standardization when:

• Data has different scales

• Using algorithms like Logistic Regression, KNN, SVM, PCA

from [Link] import StandardScaler

25
 scaler = StandardScaler()
scaled_values = scaler.fit_transform(df[["study_hours", "marks"]])

scaled_df = [Link](scaled_values, columns=["study_hours_scaled",

"marks_scaled"])
print(scaled_df)

39. Normalization
Normalization usually scales data between 0 and 1.

Formula:

X_scaled = (X - Min) / (Max - Min)

Use normalization when:

• You want values in a fixed range

• Using distance-based models or neural networks

from [Link] import MinMaxScaler

scaler = MinMaxScaler()
normalized_values = scaler.fit_transform(df[["study_hours", "marks"]])

normalized_df = [Link](normalized_values, columns=["study_hours_norm",

"marks_norm"])
print(normalized_df)

40. Simple Linear Regression Statistics

Linear regression tries to find a relationship between input X and output y.

Example:

Predict marks using study hours.

from sklearn.linear_model import LinearRegression

X = df[["study_hours"]]

26
 y = df["marks"]

model = LinearRegression()
[Link](X, y)

print("Intercept:", model.intercept_)
print("Slope:", model.coef_[0])

Equation:

marks = intercept + slope * study_hours

Prediction:

predicted_marks = [Link]([[6]])
print(predicted_marks)

41. R-Squared
R-squared measures how much variation in the target variable is explained by the model.

r2_score = [Link](X, y)
print(r2_score)

Interpretation:

If R-squared = 0.80, it means the model explains 80% of the variation in the target variable.

42. Important Statistical Tests Summary

Test Data Type Used For Python Function

Compare sample mean with

Z-Test Numerical ztest()
fixed value

One Sample T- Compare sample mean with

Numerical ttest_1samp()
Test fixed value

27
 Test Data Type Used For Python Function

Independent T- Compare means of two

Numerical + 2 groups ttest_ind()
Test groups

Numerical before/
Paired T-Test Compare same group twice ttest_rel()
after

Numerical + 3 or more Compare means of multiple

ANOVA f_oneway()
groups groups

Relationship between
Chi-Square Categorical chi2_contingency()
categories

Pearson
Numerical Linear relationship pearsonr()
Correlation

Shapiro Test Numerical Normality check shapiro()

43. Which Test Should You Use?

Numerical column vs fixed value
Use:

• Z-test if sample is large

• T-test if sample is small or population standard deviation is unknown

Example:

Average app size vs 25 MB.

Numerical column vs two categories

Use independent t-test.

Example:

Average marks of male vs female students.

Numerical column vs more than two categories

Use ANOVA.

Example:

28
Average marks across Online, Offline, and Hybrid teaching methods.

Categorical column vs categorical column

Use Chi-square test.

Example:

Gender vs Pass/Fail.

Numerical column vs numerical column

Use correlation.

Example:

Study hours vs marks.

44. Complete Mini Statistics Project

import pandas as pd
import numpy as np
import [Link] as plt
from [Link] import ttest_ind, f_oneway, chi2_contingency, pearsonr, shapiro
from [Link] import ztest

# Dataset
students = [Link]({
"student_id": range(1, 21),
"gender": ["Male", "Female"] * 10,
"method": ["Online", "Offline", "Hybrid", "Online", "Offline"] * 4,
"study_hours": [2, 4, 3, 5, 6, 7, 4, 8, 3, 6, 5, 7, 2, 4, 6, 8, 5, 7, 3, 6],
"marks": [45, 60, 55, 70, 75, 85, 65, 90, 50, 78, 72, 88, 48, 62, 77, 92,
74, 86, 53, 80]
})

students["passed"] = [Link](students["marks"] >= 50, "Yes", "No")

print([Link]())

29
Descriptive Statistics

print([Link]())
print(students["gender"].value_counts())
print([Link]("gender")["marks"].mean())

Visualization

[Link](students["marks"], bins=10)
[Link]("Marks Distribution")
[Link]("Marks")
[Link]("Frequency")
[Link]()

Z-Test
Question:

Is the average mark greater than 65?

z_stat, p_value = ztest(students["marks"], value=65, alternative="larger")

print(z_stat, p_value)

if p_value < 0.05:

print("Reject H0: Average marks are significantly greater than 65")
else:
print("Fail to Reject H0: Not enough evidence")

Independent T-Test
Question:

Do male and female students have different average marks?

male = students[students["gender"] == "Male"]["marks"]

female = students[students["gender"] == "Female"]["marks"]

t_stat, p_value = ttest_ind(male, female, equal_var=False)

print(t_stat, p_value)

30
 if p_value < 0.05:
print("Reject H0: Male and female average marks are different")
else:
print("Fail to Reject H0: No significant difference")

ANOVA
Question:

Do teaching methods have different average marks?

online = students[students["method"] == "Online"]["marks"]

offline = students[students["method"] == "Offline"]["marks"]
hybrid = students[students["method"] == "Hybrid"]["marks"]

f_stat, p_value = f_oneway(online, offline, hybrid)

print(f_stat, p_value)

if p_value < 0.05:

print("Reject H0: At least one teaching method has different average marks")
else:
print("Fail to Reject H0: No significant difference")

Chi-Square Test
Question:

Is gender related to pass/fail result?

contingency_table = [Link](students["gender"], students["passed"])

chi2_stat, p_value, dof, expected = chi2_contingency(contingency_table)

print(chi2_stat, p_value)

if p_value < 0.05:

print("Reject H0: Gender and pass/fail result are related")
else:
print("Fail to Reject H0: No significant relationship")

31
Correlation Test
Question:

Is study hours related to marks?

corr, p_value = pearsonr(students["study_hours"], students["marks"])

print(corr, p_value)

if p_value < 0.05:

print("Reject H0: Study hours and marks are significantly related")
else:
print("Fail to Reject H0: No significant relationship")

Normality Test

stat, p_value = shapiro(students["marks"])

print(stat, p_value)

if p_value < 0.05:

print("Marks are not normally distributed")
else:
print("Marks look normally distributed")

45. Common Mistakes in Statistics

Mistake 1: Using mean when data has outliers
Use median instead.

Mistake 2: Saying correlation means causation

Correlation does not prove causation.

Example:

Ice cream sales and drowning cases may both increase in summer, but ice cream does not cause drowning.

32
Mistake 3: Using t-test for more than two groups
Use ANOVA for more than two groups.

Mistake 4: Using chi-square for numerical data

Chi-square is for categorical data.

Mistake 5: Thinking p-value tells the probability that H0 is true

P-value does not directly tell that H0 is true or false.

It tells how likely the observed result is assuming H0 is true.

Mistake 6: Ignoring sample size

Very small samples can produce unreliable results.

Very large samples can make tiny differences statistically significant.

46. Statistics Roadmap for Data Science

Learn in this order:

1. Mean, median, mode

2. Variance and standard deviation
3. Percentiles, quartiles, IQR
4. Outlier detection
5. Probability basics
6. Probability distributions
7. Normal distribution
8. Z-score
9. Central Limit Theorem
10. Sampling
11. Confidence intervals
12. Hypothesis testing
13. Z-test
14. T-test
15. ANOVA
16. Chi-square test
17. Correlation
18. Regression basics
19. Statistical assumptions

33
 20. Practical interpretation

47. Final Cheat Sheet

Center of Data

df["column"].mean()
df["column"].median()
df["column"].mode()

Spread of Data

df["column"].var()
df["column"].std()
df["column"].max() - df["column"].min()

Quartiles

df["column"].quantile(0.25)
df["column"].quantile(0.50)
df["column"].quantile(0.75)

Correlation

df[["col1", "col2"]].corr()

Group Statistics

[Link]("category_column")["numeric_column"].mean()

Z-Test

from [Link] import ztest

z_stat, p_value = ztest(data, value=known_mean)

34
T-Test

from [Link] import ttest_1samp, ttest_ind, ttest_rel

ANOVA

from [Link] import f_oneway

Chi-Square

from [Link] import chi2_contingency

Pearson Correlation Test

from [Link] import pearsonr

48. Best Practical Advice

Statistics is not about memorizing formulas only.

For Data Science, always ask:

1. What type of columns do I have?

2. Am I comparing averages or categories?
3. How many groups are there?
4. Is the data numerical or categorical?
5. What is my null hypothesis?
6. What is my alternative hypothesis?
7. What does the p-value say?
8. What is the business conclusion?

That is how statistics becomes useful in real projects.

35