Hrishikesh Ghosh

Types of Data:

1. Qualitative (Categorical) Data:

o Nominal: Categories with no specific order (e.g., gender, hair color).
o Ordinal: Categories with a meaningful order, but the differences between values
are not known (e.g., satisfaction rating).
2. Quantitative (Numerical) Data:
o Discrete: Countable values (e.g., number of students in a class).
o Continuous: Measurable values that can take any number in a range (e.g., height,
weight).

Data Visualization:

1. Tables: Organized data in rows and columns to make it easier to understand.

2. Graphs and Charts:
o Bar Charts: Compare different categories.
o Pie Charts: Show proportions in a whole.
o Line Graphs: Visualize trends over time.
o Scatter Plots: Show the relationship between two continuous variables.
3. Histograms: Used to represent the distribution of numerical data, often showing the
frequency of data points within certain ranges.
4. Frequency Distributions: A summary of how often each value in a dataset occurs.

23MCAA19
Hrishikesh Ghosh

Correlation:

Correlation measures the degree to which two variables move in relation to each other.

 Pearson Correlation Coefficient (r): The most common type of correlation, it measures the
linear relationship between two variables. It ranges from -1 to 1:
o r = 1: Perfect positive correlation (as one variable increases, the other increases).
o r = -1: Perfect negative correlation (as one variable increases, the other decreases).
o r = 0: No correlation.
 Spearman Rank Correlation: Used for ranked or ordinal data, measuring the strength of a
monotonic relationship.

Regression Models:

Regression models estimate the relationship between a dependent variable (response) and one or
more independent variables (predictors).

Simple Linear Regression:

 Purpose: To model the relationship between two variables by fitting a straight line (linear
equation) to the data points.
 Equation: Y=b0+b1XY = b_0 + b_1XY=b0+b1X where YYY is the dependent variable, XXX is
the independent variable, b0b_0b0 is the intercept, and b1b_1b1 is the slope.
 Goal: Find the best-fit line that minimizes the difference (errors) between predicted and observed
values.
 Example: Predicting house prices based on square footage.

Coefficient of Determination (R²):

 Definition: A measure of how well the regression line fits the data.
 Range: 0 to 1.

23MCAA19
Hrishikesh Ghosh

o R² = 1: Perfect fit; the model explains all the variance in the dependent variable.
o R² = 0: The model explains none of the variance.
 Purpose: To quantify how much of the variability in the dependent variable can be explained by
the independent variable(s).

Significance Tests (p-value):

 Purpose: To determine if the relationship between the independent and dependent variables is
statistically significant.
 Hypothesis:
o Null Hypothesis (H₀): There is no relationship (slope = 0).
o Alternative Hypothesis (H₁): There is a significant relationship (slope ≠ 0).
 p-value: If the p-value is less than the significance level (e.g., 0.05), you reject the null
hypothesis, indicating a significant relationship.

Multiple Linear Regression:

 Purpose: To model the relationship between one dependent variable and multiple independent
variables.
 Equation: Y=b0+b1X1+b2X2+⋯+bnXnY = b_0 + b_1X_1 + b_2X_2 + \dots + b_nX_nY=b0
+b1X1+b2X2+⋯+bnXn
o Where YYY is the dependent variable, and X1,X2,…XnX_1, X_2, \dots X_nX1,X2
,…Xn are independent variables.
 Goal: Examine how each independent variable contributes to predicting the dependent variable.
 Example: Predicting house prices based on square footage, number of bedrooms, and location.

23MCAA19
Hrishikesh Ghosh

Non-Linear Regression:
 Purpose: To model relationships between variables that do not follow a straight line.
 Types:
o Exponential Regression: Models growth or decay processes (e.g., population growth,
radioactive decay).
o Polynomial Regression: Models curved relationships (e.g., U-shaped data).
 Equation: The equation depends on the specific non-linear model being used. For polynomial
regression, it could look like: Y=b0+b1X+b2X2+⋯+bnXnY = b_0 + b_1X + b_2X^2 + \dots +
b_nX^nY=b0+b1X+b2X2+⋯+bnXn
 Example: Predicting the spread of a virus over time.

23MCAA19
Hrishikesh Ghosh

Logistic Regression:
 Purpose: Used when the dependent variable is categorical (often binary, such as yes/no,
success/failure).
 Equation: log⁡(p1−p)=b0+b1X1+b2X2+⋯+bnXn\log\left(\frac{p}{1 - p}\right) = b_0 +
b_1X_1 + b_2X_2 + \dots + b_nX_nlog(1−pp)=b0+b1X1+b2X2+⋯+bnXn where ppp is the
probability of the dependent variable being 1 (success).
 Goal: Estimate the probability that the dependent variable falls into one category, based on the
independent variables.
 Example: Predicting whether a customer will buy a product (yes/no) based on their age and
income.

23MCAA19
Hrishikesh Ghosh

Discriminant Analysis:
 Purpose: A classification technique used to predict group membership (e.g., classifying
observations into predefined categories).
 Types:
o Linear Discriminant Analysis (LDA): Assumes linear boundaries between classes.
o Quadratic Discriminant Analysis (QDA): Allows for non-linear boundaries.
 Goal: Find a set of predictor variables that can best distinguish between the classes.
 Example: Predicting whether a tumor is benign or malignant based on features like size and
shape.

23MCAA19
Hrishikesh Ghosh

Factor Analysis:
 Purpose: A technique used to reduce a large number of variables into a smaller number of
underlying factors.
 Types:
o Exploratory Factor Analysis (EFA): Used to explore the underlying structure of the
data.
o Confirmatory Factor Analysis (CFA): Used to test hypotheses about the structure of
the data.
 Goal: Identify latent variables (factors) that explain the correlations among observed variables.
 Example: In psychological testing, reducing various observed personality traits into underlying
factors like extroversion or conscientiousness.

23MCAA19