One way to think of residual plots is that we turn the model—the line or curve— into a new axis.A graph from Heating Water (page 48) with a residual plot.Note: The student pages do not tell students how to zoom in. We describe one way here; you can also encourage students to look in Help. Heating Water—Teacher NotesThis activity is the second of three that introduce students to residuals and residual plots. Our approach in Residuals Playground and Heating Water is to have students calculate residuals directly—by having Fathom calculate from the given function—before introducing the Make Residual Plot command.The third activity, Radiosonde Residuals, is suitable either as practice to follow the second or as a quick introduction for more experienced students. It’s also suitable for homework or open lab time.What’s Important HereWe use residuals for many purposes in data analysis. These activities focus on using residuals to see patterns that are invisible in a “regular” graph.• You can see deviations (from the curve) or a pattern (e.g., nonlinearity) better because the vertical scale of the residual plot is so different from the one in the main plot. Making the residual plot magnifies the data with respect to the model (the line).• If residuals don’t look random, we should question the model. In Heating Water, for example, the linear model is OK for some purposes (interpolating temperatures to within a degree, say), but we can see that the phenomenon is really not linear.• Discussion Questions1. What are the units on the vertical scale of the residual plot? (degrees Celsius)2. Compare the main plot to the residual plot. Why are the scales so different?3. (Especially in Residuals Playground.) What would happen if we defined a residual as (predicted – observed) instead of the other way around? (For one thing, when we moved the data one way, residuals would go the other. It would be confusing.)Tech Notes• Students need to zoom in for Heating Water. To do so, hold down Ctrl (Win) Option (Mac) so that the cursor changes to a magnifying glass and drag a rectangle around the region you want to investigate. The rectangle expands to fill the whole graph. Rechoosing Scatter Plot from the popup menu in the graph rescales the plot.• You can do Heating Water with your own data (especially if you’re teaching chemistry, for example). If the burner is hot, you get a good, linear-looking regime in the middle. At the beginning, it flattens out as the container heats; at the end, it flattens as the water gets closer to boiling. If you include these “flat” areas, (a) students should definitely use the movable line—not least-squares—and fit it to the linear part of the data, and (b) students will need to expand the vertical scale of the residual plot in order to see the curve in the linear part of the temperature graph.
đang được dịch, vui lòng đợi..
