6西格玛回归概论.ppt

* Explanation The storks were nesting on chimneys in town versus nesting sites in rural areas. The storks were attracted to higher chimney temperatures because they aided in the incubation of their eggs while they were nesting, therefore the number of stalks do not necessarily affect the population but the size of the population (and therefore number of chimneys) may well affect the number of stalks. This example taken from Box, Hunter and Hunter - Statistics for experimenters, page 8. * Caution: Eventually the car may become a “classic” * Draw Curve onto plot The value may follow a “bath tub” curve over time * Single Y with multiple X’s Point this page out in the Pocket Guide * 2 or more x’s with a single Y No chart or best fit line is created for this tool. We will use the session window output to review the results. * Talk about the single Y and the multiple X’s This should be a condensed list to arrive at the vital few. Please note that this is a great Admin example * Go to the MultRegr.MPJ project Open Sales.MTW Walk the room to make sure that everyone is caught up! * Try to keep the regression model as simple as possible. Remember the vital few Refer back to the Y is a function of X’s * Between Collection and Analysis phases of the multiple regression roadmap we should do best subsets analysis and perform correlation analysis to see if any collinearity / confounding is present. * Collinearity reflects the confounding that may be present between the individual X variables. Some of the effects are confounded with each other. That is, you cannot estimate the magnitude of the effect of the individual Xs separately. Good Example is shoe Size, height and weight of 11 year olds. All of these may be going up together at the same time. * Ask the class to go back to the MultRegr.MPJ and the worksheet Sales.MTW file * Minitab path Stat Basic statisticsCorrelation Enter the X variables that are shown on the screen. Make sure they uncheck the P