SJ
Premium Platinum
It's been a few years since I ran a regression, but I have some undergraduate background in it.
With the big Geelong-Essendon crowd predicted at Etihad this Friday night, I was thinking about which factors influence the crowd size, how one would best measure them, and how one would best incorporate them into an expression.
So, given games that:
- feature Geelong, and
- feature Essendon, and
- are played at Etihad Stadium
Crowd size would depend on at least:
- form of both sides (recent form and/or season performance)
- evenness of both sides (measured by how close form of both sides is)
- timeslot (Friday night, Saturday day, Saturday night, Sunday day)
- home game club
- time of season (I assume this is a multiplier rather than an indepedent linear variable, will have a greater effect positively or negatively the later it is in the season)
Can anyone suggest some expressions (rather than just the linear y = ax1 + bx2 + ... + c) to run?
With the big Geelong-Essendon crowd predicted at Etihad this Friday night, I was thinking about which factors influence the crowd size, how one would best measure them, and how one would best incorporate them into an expression.
So, given games that:
- feature Geelong, and
- feature Essendon, and
- are played at Etihad Stadium
Crowd size would depend on at least:
- form of both sides (recent form and/or season performance)
- evenness of both sides (measured by how close form of both sides is)
- timeslot (Friday night, Saturday day, Saturday night, Sunday day)
- home game club
- time of season (I assume this is a multiplier rather than an indepedent linear variable, will have a greater effect positively or negatively the later it is in the season)
Can anyone suggest some expressions (rather than just the linear y = ax1 + bx2 + ... + c) to run?