Notes
Slide Show
Outline
1
APS 425
Spring 2012

  • Boston Marathon Data


  • Instructor: G. William Schwert
  • 585-275-2470
  • schwert@schwert.simon.rochester.edu
2
Winning Times in Marathon
  • Sports records are often interesting because they reflect the evolution of human development.


  • This data file contains the winning times, in seconds, for the Boston Marathon from 1897-2012.
3
Winning Times in Marathon
  • As this race has become more famous and the prize for winning has become lucrative, world-class runners have begun to compete and win, so I also include variable indicating whether the winning racers were from the U.S., or not.
  • Pro equals 1 since 1986 to reflect the payment of prize money
  • Women have participated in this race since 1966, so the winning times for women are included from 1966-2012.



4
Variables
  • m_sec (winning time for men in seconds, so two hours = 60*60*2 = 7200 seconds)
  • Note that the wining time in 1918 is not available because of WW I
  • w_sec (winning time for women in seconds, since 1966)
  • time  (a time trend = -53 in 1897 and =62 in 2012)
5
Variables
  • usm  (dummy variable = 1 when the male winner is from the US, and 0 otherwise)
  • Note that I have set this = 0 from 2011-2050 on the assumption that foreign runners will continue to win in the future)
  • usw  (dummy variable = 1 when the female winner is from the US, and 0 otherwise)
  • Note that I have set this = 0 from 2011-2050 on the assumption that foreign runners will continue to win in the future)


  • pro  (dummy variable = 1 since 1986, and 0 otherwise
6
Variables
  • Weather_bad  if commentary on history web site describes conditions as unfavorable (hot, headwind, or very cold and rainy, for example)
  • Weather_good  if commentary on history web site describes conditions as favorable (e.g., strong tailwind)
7
Exponential Trend Model for
Male Winning Times
8
Residual Autocorrelations
9
Exponential Trend Model for
Male Winning Times, AR(3)
10
Actual & Residual Plot
11
Forecasts for 2010-2050
12
Exponential Trend Model for
Female Winning Times, AR(1)
13
Actual & Residual Plot
14
Female Winning Times Related to Male Winning Times, AR(1)
15
Forecasts for 2010-2050
16
Links

  • Eviews worksheet
  • http://schwert.simon.rochester.edu/a425/a425_boston.wf1


  • APS 425 Home Page
  • http://schwert.simon.rochester.edu/a425/a425main.htm