ECIV 301 Programming & Graphics Numerical Methods for Engineers Lecture 26 Regression Analysis-Chapter 17
Curve Fitting Often we are faced with the problem… what value of y corresponds to x=0.935?
Curve Fitting Question 2 : Is it possible to find a simple and convenient formula that represents data approximately ? e.g. Best Fit ? Approximation
Experimental Measurements Strain Stress
Experimental Measurements Strain Stress
BEST FIT CRITERIA Strain y Stress Error at each Point
Best Fit => Minimize Error Best Strategy
Best Fit => Minimize Error Objective: What are the values of a o and a 1 that minimize ?
Least Square Approximation In our case Since x i and y i are known from given data
Least Square Approximation 2 Eqtns 2 Unknowns
Least Square Approximation
Example
Quantification of Error Average
Quantification of Error Average
Quantification of Error Average
Quantification of Error Standard Deviation Shows Spread Around mean Value
Quantification of Error
“Standard Deviation” for Linear Regression
Quantification of Error Better Representation Less Spread
Quantification of Error Coefficient of Determination Correlation Coefficient
Linearized Regression The Exponential Equation
Linearized Regression The Power Equation
Linearized Regression The Saturation-Growth-Rate Equation
Polynomial Regression A Parabola is Preferable
Polynomial Regression Minimize
Polynomial Regression
3 Eqtns 3 Unknowns
Polynomial Regression Use any of the Methods we Learned
Polynomial Regression With a 0, a 1, a 2 known the Total Error Standard Error Coefficient of Determination
Polynomial Regression For Polynomial of Order m Standard Error Coefficient of Determination