Prediction (Optional) · Exercise 15

contains real data for the first two decades of flu reporting.
$$\begin{array}{ccc}\text{Year}& \text{Number of Flu Cases Diagnosed}& \text{Number of Flu Deaths}\\ \text{Pre-1981}& \text{91}& \text{29}\\ \text{1981}& \text{319}& \text{121}\\ \text{1982}& \text{1,170}& \text{453}\\ \text{1983}& \text{3,076}& \text{1,482}\\ \text{1984}& \text{6,240}& \text{3,466}\\ \text{1985}& \text{11,776}& \text{6,878}\\ \text{1986}& \text{19,032}& \text{11,987}\\ \text{1987}& \text{28,564}& \text{16,162}\\ \text{1988}& \text{35,447}& \text{20,868}\\ \text{1989}& \text{42,674}& \text{27,591}\\ \text{1990}& \text{48,634}& \text{31,335}\\ \text{1991}& \text{59,660}& \text{36,560}\\ \text{1992}& \text{78,530}& \text{41,055}\\ \text{1993}& \text{78,834}& \text{44,730}\\ \text{1994}& \text{71,874}& \text{49,095}\\ \text{1995}& \text{68,505}& \text{49,456}\\ \text{1996}& \text{59,347}& \text{38,510}\\ \text{1997}& \text{47,149}& \text{20,736}\\ \text{1998}& \text{38,393}& \text{19,005}\\ \text{1999}& \text{25,174}& \text{18,454}\\ \text{2000}& \text{25,522}& \text{17,347}\\ \text{2001}& \text{25,643}& \text{17,402}\\ \text{2002}& \text{26,464}& \text{16,371}\\ \text{Total}& \text{802,118}& \text{489,093}\end{array}$$

Do you think a linear fit is best? Why or why not?

Its source prints no answer for this problem, so there's nothing to check against.

OpenStax, Statistics · §12.4 Prediction (Optional) · Exercise 15