The LSRL for predicting weight (y, kg) from height (x, cm) is y^=−100+0.8x. Interpret the slope and intercept, predict weight for height=175 cm, and explain why extrapolating to height=50 cm is problematic.
Example 3
medium
Given xˉ=4, yˉ=20, r=0.8, sx=2, sy=5, find the LSRL.
Example 4
medium
Why is predicting y at an x-value far outside the observed range dangerous? Give one example.
Example 5
hard
If y is shifted by adding 5 to each observation, how does the LSRL change?
Practice Problems
Try each one, then open the solution to compare your method.
Example 1
easy
Given y^=5+3x: (a) predict y when x=4, (b) interpret the slope, (c) does the line pass through the origin?
Example 2
hard
The LSRL has the property of minimizing ∑ei2=∑(yi−y^i)2. Explain why minimizing squared residuals (rather than absolute residuals) is preferred, and name two consequences of this choice.
Example 3
easy
A regression line is y^=3+2x. Predict y when x=4.
Example 4
easy
In y^=5+3x, what is the slope?
Example 5
easy
In y^=7−2x, what is the y-intercept?
Example 6
easy
The LSRL minimizes the sum of squared what?
Example 7
easy
Compute the slope: r=0.8, sy=4, sx=2. Use b=rsxsy.
Example 8
easy
A line passes through the point of averages (xˉ,yˉ)=(5,12) with slope b=2. Find the intercept using a=yˉ−bxˉ.
Example 9
easy
In y^=10+4x where y is cost in dollars and x is hours, interpret the slope.
Example 10
easy
Does the regression of y on x generally give the same line as the regression of x on y?
Example 11
medium
Given r=0.6, sx=5, sy=10, xˉ=20, yˉ=50, find the LSRL.
Example 12
medium
Using y^=26+1.2x, predict y at x=30.
Example 13
medium
Two data points lie exactly on y^=2+3x: (1,?) and (4,?). Find both predicted values.
Example 14
medium
A model y^=100+5x predicts plant height (cm) from days x. Why is predicting height at x=10,000 days unwise?
Example 15
medium
Find the slope of the LSRL through summary stats r=−0.5, sy=8, sx=4.
Example 16
medium
In y^=200−0.5x, y is weight (lb) and x is age in days for a dieting program. Interpret the intercept and say whether it is meaningful.
Example 17
medium
A slope is computed as b=rsxsy with b=3 and sxsy=5. Find r.
Example 18
medium
The LSRL is y^=4+2x. A new point (3,12) is observed. Is the observed value above or below the line?
Example 19
medium
A line passes through (xˉ,yˉ)=(8,30) with slope b=2.5. Find its equation.
Example 20
challenge
A dataset has xˉ=10,yˉ=50,sx=2,sy=6,r=0.9. Find the LSRL and predict y at x=12.
Example 21
challenge
Show that if r=1, the regression of y on x and of x on y describe the same line. (Use slopes byx=rsxsy and bxy=rsysx.)
Example 22
challenge
A regression on temperature (x, in ∘C) gives y^=2+0.5x. If temperature is re-expressed in tenths of a degree (x′=10x), what is the new slope?
Example 23
easy
Given y^=8+1.5x, predict y when x=6.
Example 24
easy
Given r=0.5, sy=6, sx=3, compute the slope of the LSRL.
Example 25
easy
For LSRL y^=100−2x predicting weight (lb) from days x in a diet program, interpret the slope.
Example 26
easy
Residual = observed − predicted. If yobs=15 and y^=12, find the residual.
Example 27
medium
A regression y^=50−0.2x predicts time-on-test (min) from age x (yr). Predict for x=25 and compute residual if observed time is 42.
Example 28
medium
Given the LSRL has slope b=0.6 and passes through (xˉ,yˉ)=(12,25), find the equation.
Example 29
medium
In a regression with r=0.9, r2 describes what?
Example 30
medium
Compute slope: r=−0.6, sy=12, sx=4.
Example 31
medium
A residual plot shows a clear curved pattern. What does this indicate?
Example 32
medium
For y^=30+4x, predict the change in y if x increases by 5.
Example 33
medium
Given five data points (1,3),(2,5),(3,7),(4,8),(5,10), compute xˉ and yˉ.
Example 34
medium
For the LSRL passing through (xˉ,yˉ)=(3,6.6) with slope 1.7, write the equation.
Example 35
hard
For y^=20+5x in dollars per item (y) vs. quantity (x), interpret slope and intercept and judge if intercept is meaningful when x=0.
Example 36
hard
A regression equation rescales x to x′=x/10. If original LSRL is y^=5+2x, what is the new equation in x′?
Example 37
hard
For the LSRL y^=4+2x, the sum of residuals over all data points is always what?
Example 38
hard
A regression has slope b=3. If y is rescaled to y′=2y, what is the new slope?
Example 39
hard
For r=0.6, what percent of variation in y is NOT explained by linear regression on x?
Example 40
challenge
Given that the LSRL of y on x has slope byx and the LSRL of x on y has slope bxy, show byx⋅bxy=r2.
Example 41
challenge
A regression line y^=3+2x has r2=0.81 and sx=4. Find sy.
What to focus on
Core idea:The least-squares regression line is the unique line y^=a+bx minimizing the total squared vertical distance to the data.
Common stuck point:The procedure for least squares regression line is the easy part; the trap is minimizing perpendicular or horizontal distances. Asking "Am I fitting a single straight line to two-variable numeric data by minimizing squared vertical distances?" first is what keeps a correct-looking calculation from being attached to the wrong concept.
Ask yourself:Am I fitting a single straight line to two-variable numeric data by minimizing squared vertical distances?
Concept recap
The unique straight line y^=a+bx that minimizes the sum of squared vertical distances (residuals) between the observed data points and the line.
Imagine each data point attached to a stiff rod by a vertical spring whose pull grows with its stretch. Let go, and the rod settles where all the pulls balance: close to most points, tugged hardest by the farthest. That resting position is the least-squares line. Like any rod, it stays straight, so it can't follow a curved pattern.