BME Vegyészmérnöki és Biomérnöki Kar

Nyomtatás

Design of Experiments

BMEVEKFM203

Statistical intervals. Confidence, tolerance and prediction intervals, their use and interpretation. Error propagation (measurement uncertainty) analysis. Study of error sources of analytical measurements according to the EURACHEM Guide. Regression analysis. Parameter estimation using least squares for fitting a straight line. Assumptions and checking their validity, adequacy of the function fitted, plotting the residuals. Methods applied when the variance is not constant (weighted least squares and maximum likelihood). The calibration problem:confidence interval for the unknown value of the independent variable. Multiple linear regression. Nonlinear regression. Analysis of variance: One-way ANOVA, multiple comparisons (contrasts). Assumptions and checking their validity. Cross-classification. Box-Cox transformation. Advanced ANOVA: random factors, nested design, mixed model, random block. Design of factorial experiments Two-level full and fractional designs, Experimental optimisation using gradient and simplex method. Quadratic designs: 3k, Box-Behnken and central composite designs. Statistical methods for method validation Repeatability and reproducibility, robustness, linearity. The seminars contain discussion of case studies and solving simpler exercises, students are requested to solve problems of medium complexity alone using statistical software (Statistica).