STAT 101 Lecture Notes - Lecture 5: Mathematical Notation, Statistical Model, Standard Deviation

• School: Simon Fraser University
• Department: Statistics
• Course: STAT 101
• Professor: Timothy Swartz

Document Summary

Mathematical notation: y = mx + b; m= slope, b = intercept. Statistical notation: y hat = bx + a; a = intercept, b = slope, h hat = predicted y value. Slope tells us the change in y for every one-unit increase in x. A good statistical model (regression line) has small prediction errors (called residuals) Prediction error (residual) = observed y predicted y-hat. Squared error = residual ^ 2 = (y y-hat) ^2. The least-squares regreesion line is the line with the smallest sse. R =correlation coefficient s= standard deviation (note: find slope before intercept) Correlation: measures direction and strength of linear association between x and y. No explicit distinction between explanatory variable and response variable. Doesn"t matter which one is x, which one is y. X and y can be interchanged in correlation, but not in the formula for a regression line: y =a+bx.