The document discusses linear correlation and linear regression. It defines correlation as measuring the relationship between paired variables and identifies three degrees of correlation: perfect, some, and no correlation. It provides examples of correlated variables and measures of correlation using Pearson's r correlation coefficient. Linear regression finds the linear relationship between a dependent and independent variable and predicts values using the regression line equation. The standard error of estimate measures how far data points deviate from the regression line. Multiple regression allows predicting a dependent variable from multiple independent variables using a multiple regression equation.
The document discusses linear correlation and linear regression. It defines correlation as measuring the relationship between paired variables and identifies three degrees of correlation: perfect, some, and no correlation. It provides examples of correlated variables and measures of correlation using Pearson's r correlation coefficient. Linear regression finds the linear relationship between a dependent and independent variable and predicts values using the regression line equation. The standard error of estimate measures how far data points deviate from the regression line. Multiple regression allows predicting a dependent variable from multiple independent variables using a multiple regression equation.
The document discusses linear correlation and linear regression. It defines correlation as measuring the relationship between paired variables and identifies three degrees of correlation: perfect, some, and no correlation. It provides examples of correlated variables and measures of correlation using Pearson's r correlation coefficient. Linear regression finds the linear relationship between a dependent and independent variable and predicts values using the regression line equation. The standard error of estimate measures how far data points deviate from the regression line. Multiple regression allows predicting a dependent variable from multiple independent variables using a multiple regression equation.
The document discusses linear correlation and linear regression. It defines correlation as measuring the relationship between paired variables and identifies three degrees of correlation: perfect, some, and no correlation. It provides examples of correlated variables and measures of correlation using Pearson's r correlation coefficient. Linear regression finds the linear relationship between a dependent and independent variable and predicts values using the regression line equation. The standard error of estimate measures how far data points deviate from the regression line. Multiple regression allows predicting a dependent variable from multiple independent variables using a multiple regression equation.
Linear Regression A. Linear Correlation ⚫ Correlation – measures the degree and direction of the relationship between paired variables. ⚫ 3 Degrees of Correlation ⚫ 1. perfect (±) correlations ⚫ 2. some degree of (±) correlation ⚫ 3. no correlation Degrees of Correlation Interpretation Examples of Correlation: ⚫ Weight vs. Height (tall people tend to be heavier than small people) ⚫ GNP vs. Aggregate Investment ⚫ Work Output vs. Work Experience ⚫ Statistics vs. Technical Writing ⚫ Computer Literacy vs. Personality ⚫ Wine Consumption vs. Illness ⚫ Students’ Grades vs. # of Hours spent in Studying Measures of Correlation Pearson r Correlation
r= N(∑XY) – (∑X)(∑Y) √[N(∑X2) – (∑X)2][N(∑Y2) – (∑Y)2 Extent of Correlation Between Paired Variables ⚫ 0.00 - ±0.09 - No Correlation ⚫ ±0.10 - ±0.20 - Negligible Correlation ⚫ ±0.21 - ±0.40 - Low Correlation ⚫ ±0.41 - ±0.60 - Substantial Correlation ⚫ ±0.61 - ±0.80 - Marked Correlation ⚫ ±0.81 - ±0.99 - High to Very High Corr. ⚫ ±1.00 - Perfect Correlation Example: ⚫ Table 1 presents data relating the number of weeks of experience in a job involving the wiring of miniature electronic components and the number of components which were rejected during the past week for 12 randomly selected workers. ⚫ i) Compute the correlation coefficient for the data. ⚫ ii) Test if the correlation is significant. Table 1: Weeks of Experience & Number of Components Rejected During a Sampled Week for 12 Assembly Workers
Sampled 1 2 3 4 5 6 7 8 9 10 11 12 Worker
Wks. Of Experie 7 9 6 14 8 12 10 4 2 11 1 8 nce (X)
# of Rejects 26 20 28 16 23 18 24 26 38 22 32 25 (Y) W X Y X2 y2 XY 1 7 26 49 676 182 2 9 20 81 400 180 3 6 28 36 784 168 4 14 16 196 256 224 5 8 23 64 529 184 6 12 18 144 324 216 7 10 24 100 576 240 8 4 26 16 676 104 9 2 38 4 1444 76 10 11 22 121 484 242 11 1 32 1 1024 32 12 8 25 64 625 200 ∑X=92 ∑Y=298 ∑X2=876 ∑Y2=7798 ∑XY=2048 Solutions: ⚫ i.1. Using Pearson (r) r= 12(2,048) – (92)(298) [12(876)–922][12(7,598)–2982] r = - 0.908 ≈ 0.91 very strong negative correlation ii. Test for Significance ⚫ Ho: r = 0 Ha: r ≠ 0 ⚫ ά(.05) = 0.5760 ά(.01) = 0.7079 ⚫ df = n – 2 = 12 – 2 = 10 ⚫ rc rt ⚫ - 0.908 > - 0.7079 , reject Ho ⚫ The computed r is significant, which implies that a very strong negative correlation exist between the 2 variables. The (-) correlation means that the longer the work experience in a job involving the wiring of miature component, the lesser the number of rejected components. B. Linear Regression
⚫- is a technique used to approximate or predict the values or relationship between dependent and independent (predictor) variable. Regression Techniques ⚫ 1. Scatter Diagram – visual or graphical method of determining the relationship of paired data. ⚫ 2. Equation of the Regression Line (least-squares line) – is used as a predictor of the y-values. ⚫ 3. Standard Error of Estimate (Se) – measures the amount of spread of the sample points about the regression line. Formulas: ⚫ Equation of the Regression Line ⚫ Ypred. = mx + b ⚫ b = Ῡ - mX Ῡ = ∑Y/n ⚫ m = n(∑XY) - ∑X∑Y X = ∑X/n ⚫ n∑X2 – (∑X)2 ⚫ Standard Error of Estimate ⚫ Se = ∑Y2 – b(∑Y) – m(∑XY) ⚫ n–2 Examples: ⚫ 1. Each of the following pairs represents the number of licensed drivers (X) and the number of cars (Y) for houses in a posh village in Metro Manila. Predict the number of cars for each of 2 new families with 2 and 5 drivers. ⚫ Drivers(X) | 5 5 2 2 3 1 2 ⚫ Cars(Y) | 4 3 2 2 2 1 2 Solution:
X Y XY X2 Y2 5 4 20 25 16 5 3 15 25 9 2 2 4 4 4 2 2 4 4 4 3 2 6 9 4 1 1 1 1 1 2 2 4 4 4 ∑X= 20 ∑Y=16 ∑XY=54 ∑X2=72 ∑Y2=42 ⚫X = ∑X/N = 20/7 = 2.86 ⚫ Ῡ = ∑Y/N = 16/7 = 2.29 ⚫ m = n(∑XY) - ∑X∑Y = 7(54) – (20)(16) ⚫ n∑X2 – (∑X)2 7(72) – (20)2 ⚫ m = 0.56 ⚫ b = Ῡ - mX = 2.29 – (0.56)(2.86) = 0.69 ⚫ Ypred. = mx + b = 0.56X + 0.69 ⚫ If x=2, Ypred. = 0.56(2)+0.69 = 1.81≈ 2 ⚫ If x=5, Ypred. = 0.56(5)+0.69 = 3.49≈ 4 ⚫ 2. The following is the length of time that 10 job applicants for employment abroad have studied College and their scores in a proficiency test in English. Yrs. In College| 5 3 2 3 4 2 3 4 4 5 Scores in Eng. |84 63 48 75 78 58 57 72 73 89 a. Construct a scatter diagram. b. Find the equation of a regression line. c. Sketch the graph of the regression line. d. Compute for Se. e. If X = 4 ½ , what is Ypred. ? ⚫ a) Solutions: ⚫ 90 · ⚫ · ⚫ 80 · ⚫ · ·· ⚫ 70 ⚫ ⚫ 60 · ⚫ · · ⚫ 50 ⚫ · ⚫ 1 2 3 4 5 X Y XY X2 Y2 5 84 420 25 7056 3 63 189 9 3969 2 48 96 4 2304 3 75 225 9 5625 4 78 312 16 6084 2 58 116 4 3364 3 57 171 9 3249 4 72 288 16 5184 4 73 292 16 5329 5 89 445 25 7921 ∑X=35 ∑Y=697 ∑XY=2554 ∑X2=133 ∑Y2=50085 ⚫ b) Ῡ = 697/10 = 69.7 ⚫ X = 35/10 = 3.5 ⚫ m = 10(2554) – (35)(697) = 10.9 ⚫ 10(133) – 352 ⚫ b = 69.7 – 10.9(3.5) = 31.55 ⚫ Ypred.= 10.9x + 31.55 ⚫ e) If X= 4½, ⚫ Ypred. = 10.9(4.5)+31.55= 80.6 ≈ 81 ⚫ d) Standard Error of Estimate
⚫ Se = 50,085 – 31.55(697) – 10.9(2554)
⚫ 10 – 2
⚫ Se = 0.47 Multiple Regression Analysis
-allows you to measure how
much of the variation in a series can be attributed to another series. Multiple Regression Analysis - is used when we have several variables that we can use to predict what we’re interested in. Multiple Regression Equation Y = b0 + b1X1 + b2X2 + … + bnXn where: •Y is the dependent variable to be predicted •b0is the intercept •x1, x2, x3, ..., xnare the the known independent variables that may influence y •b1, b2, b3, ..., bnare the numerical constants which must be determined from the observed data Thank you
