TN Online TestSamacheer Kalvi practice

12th Statistics Chapter 5: Regression Analysis

12th Standard · Statistics · #5

Class 12 Statistics chapter 5 explains regression analysis: simple, multiple and non-linear regression, why there are two regression lines, fitting a line of best fit by the method of least squares, regression coefficients and their properties, and how regression differs from correlation for prediction.

Study this chapter

Asked in previous board exams

8 questions from this chapter came in 4 past public exams. Open any of them to see the answer and explanation.

  1. March 2026
    __________ is the Father of mental tests.
    See answer in the solved paper →
  2. March 2026
    The regression lines intersect at :
    See answer in the solved paper →
  3. March 2025
    If the two lines of regression are perpendicular to each other then \(r_{xy}\) = __________.
    See answer in the solved paper →
  4. March 2025
    The linear equation Y = a + bx is called as regression equation of __________.
    See answer in the solved paper →
  5. March 2023
    If \(b_{XY} = 0.7\) and 'a' = 8 then the regression equation of X on Y is :
    See answer in the solved paper →
  6. March 2023
    If the two lines of regression are perpendicular to each other, then \(r_{XY}\) =
    See answer in the solved paper →
  7. March 2020
    If \(b_{xy} > 1\) then, \(b_{yx}\) is :
    See answer in the solved paper →
  8. March 2020
    If \(\bar{x} = 7.875\), \(\bar{y} = 27.5\), \(b_{yx} = 2.098\), then the regression equation of y on x is :
    See answer in the solved paper →

About Regression Analysis

Medium ~150 min study

Regression Analysis is chapter 5 of the 12th standard Statistics textbook of the Tamil Nadu State Board. Correlation tells us the type and intensity of a linear relationship between two variables, but it gives no mathematical function for prediction. Regression analysis fills that gap: it finds the functional relationship between a dependent (response) variable and one or more independent variables (regressors or predictors).

The chapter opens with Sir Francis Galton, who coined the terms correlation and regression while studying how extreme characteristics such as height are not fully passed on from parents to offspring. It then defines simple linear, multiple linear and non-linear regression, explains why two regression lines exist, and fits the line of best fit by the method of least squares through the normal equations.

The second half derives the regression coefficients \( b_{YX} \) and \( b_{XY} \), links them with Karl Pearson's correlation coefficient, lists their properties and the properties of regression equations, works ten examples and closes with the differences between correlation and regression. The practice set here has the 20 book-back multiple choice questions and 15 additional questions, most of them short calculations.

What you'll learn

  • Define regression and distinguish simple linear, multiple linear and non-linear regression.
  • Explain why two regression lines exist and when only one is meaningful.
  • Fit a simple linear regression equation by solving the normal equations of least squares.
  • Calculate b_YX and b_XY from raw sums, deviations or standard deviations and correlation.
  • Use the slope-point form to estimate one variable from a given value of the other.
  • Apply the properties of regression coefficients and lines, and contrast regression with correlation.

Before you start

  • Mean, variance and standard deviation of a variable.
  • Karl Pearson's correlation coefficient and covariance from chapter 4.
  • Equation of a straight line in slope-intercept and slope-point form, and solving two linear equations.

Topics covered in this chapter

Definition of Regression A statistical method of finding the mathematical functional relationship between a dependent variable and one or more independent variables, mainly for prediction.
Types of Regression Simple linear regression with one regressor, multiple linear regression with two or more regressors, and non-linear regression when the relationship is not a straight line.
Two Regression Lines The regression line of Y on X and the line of X on Y, both valid when the variables are interchangeable; only one line when X alone causes Y.
Method of Least Squares Fits the line of best fit by making the sum of residuals zero and the sum of squared residuals the least, leading to the two normal equations.
Regression Coefficients The slopes b_YX and b_XY, obtained from covariance over variance or from the correlation coefficient times the ratio of standard deviations.
Properties of Regression Correlation is the geometric mean of the coefficients, both share one sign, origin change has no effect, and the lines meet at the means.
Correlation and Regression Correlation measures only the nature and extent of a relationship and is symmetric; regression studies impact, predicts, and is not symmetric.

Regression Analysis explained

What the chapter covers

Definition and types

Regression analysis is a statistical method of determining the mathematical functional relationship connecting independent variable(s) and a dependent variable. With one independent variable and a linear relationship it is simple linear regression; with two or more it is multiple linear regression; if the relationship is not linear it is non-linear regression. Multiple and non-linear regression are given only for knowledge.

Two regression lines

When X and Y are interchangeable, we get the regression line of Y on X and the line of X on Y. If X alone causes Y, there is only the line of Y on X.

Method of least squares

The line of best fit makes the sum of the residuals zero and the sum of their squares the least. Minimising \( E(a,b) = \sum (y_i - a - bx_i)^2 \) gives the normal equations \( na + b\sum x_i = \sum y_i \) and \( a\sum x_i + b\sum x_i^2 = \sum x_i y_i \), so \( \hat{a} = \bar{y} - \hat{b}\bar{x} \) and \( \hat{b} = \frac{Cov(X,Y)}{V(X)} \).

Regression coefficients

\( b_{YX} = r_{XY} \frac{SD(Y)}{SD(X)} \) and \( b_{XY} = r_{XY} \frac{SD(X)}{SD(Y)} \). Slope-point forms: \( \hat{Y} - \bar{y} = b_{YX}(x - \bar{x}) \) and \( \hat{X} - \bar{x} = b_{XY}(y - \bar{y}) \).

Properties

  • \( r_{XY} = \sqrt{b_{XY} \times b_{YX}} \), the geometric mean, with the same sign as the coefficients.
  • If one coefficient exceeds 1, the other is below 1; their arithmetic mean is at least \( r_{XY} \).
  • Independent of change of origin but not of scale.
  • If \( r = 0 \) the lines are perpendicular; if \( r = \pm 1 \) they coincide; they intersect at \( (\bar{X}, \bar{Y}) \).

Correlation versus regression

Correlation is symmetric (\( r_{XY} = r_{YX} \)) and measures only the nature and extent of the relationship; regression is not symmetric and is used for prediction, only within the range of the data (interpolation).

Common mistakes to avoid

  • Swapping b_YX and b_XY. Remember that b_YX has SD(Y) on top and is the slope of the line of Y on X.
  • Taking the correlation as the arithmetic mean of the coefficients. It is the geometric mean, with their common sign.
  • Forgetting to subtract the means in the slope-point form, and so writing Y = b x instead of Y - mean of y = b (x - mean of x).
  • Using a variance where a standard deviation is needed in b = r SD(Y)/SD(X).
  • Using a fitted equation far outside the range of the data. Regression supports interpolation, not extrapolation.

Test yourself on these with the practice test, then check the worked reasoning in the solved MCQs.

Frequently asked questions

What is regression analysis?

It is a statistical method of determining the mathematical functional relationship connecting independent variables and a dependent variable. It shows how the typical value of the response changes when a regressor changes, and is widely used for prediction.

Why are there two regression lines?

When X and Y are interchangeable, either can be taken as the independent variable, giving the line of Y on X and the line of X on Y. If X alone causes Y, only the line of Y on X is used.

What does the method of least squares do?

It chooses a and b so that the sum of the residuals is zero and the sum of the squares of the residuals is the least. Differentiating gives the two normal equations, whose solution is the line of best fit.

How is the correlation coefficient found from regression coefficients?

The correlation coefficient is the geometric mean of the two regression coefficients, r = square root of b_XY times b_YX, and it carries the same sign as the coefficients.

Where do the two regression lines meet?

Both lines pass through the point of averages, so they intersect at the means of X and Y. Solving the two regression equations together therefore gives the mean values.

Are regression coefficients affected by change of origin or scale?

They are independent of the change of origin, so deviations from assumed means can be used. They are not independent of the change of scale.

How does regression differ from correlation?

Correlation shows only the nature and extent of a linear relationship and is symmetric in x and y. Regression studies the impact of the independent variable on the dependent variable, is used for prediction and is not symmetric.

Last updated 28 October 2026

More chapters in Statistics

View all
1 Tests of Significance - Basic Concepts and Large Sample Tests 2 Tests Based on Sampling Distributions - I 3 Tests Based on Sampling Distributions - II 4 Correlation Analysis 6 Index Numbers 7 Time Series and Forecasting 8 Vital Statistics and Official Statistics