- Starting with the points in their original positions, choose "manipulate a
line" and try to fit a line to the data on the graph by either dragging the
line, using the sliders, or typing.
One popular way to compute a best fit line is to minimize the sum of the
squares of the difference between the data points and the line. This approach
is known as "least squares fit."
- Turn on the "show error squares" option, note the total error, then try to
minimize it by adjusting your trial line.
- To see the actual least-squares fit line, select "compute least squares
fit line."
- How does this compare to your trial line?
- To see the least squares error, snap the trial line to the least squares
line using the button next to right of the equation of the (green) trial
line.
- Turn off the "compute least squares fit line" and "show error squares"
options. Drag the left most point to the lower left part of the graph area and
repeat steps 1 - 3 above.
- What can you say about the effect of extreme data points on the
usefulness of the least squares approach?