football-score

master project 2020

Install guide

It's assumed that C++ and R are already installed.

R dependencies:

• TMB
• optparse
• jsonlite
• Matrix
• expm
• skellam

optional dependencies:

• rstan
• tmbstan

Depending on your workspace, these may be installed automatically when you run the program.

To download the scripts, cd into your install directory and run

git clone https://github.com/2039/football-score.git

Data

The data/ folder must be filled with data before use. This should contain three files, scores.csv, teams.csv, scores_full.csv.

The header of each file is:

• scores.csv : home_team,away_team,home_team_score,away_team_score
• teams.csv : key,name
• scores_full.csv : Hjemmelag,Bortelag,Hjemmeskår,Borteskår,Runde,Dato,Dag,Tid,Bane,Kampnummer

The translation of the scores_full.csv header is : home_team,away_team,home_score,away_score,round,date,weekday,time,stadium,match_id. We note that the type of home_team and away_team differ in the tables.

The type of each row is:

• scores.csv : int,int,int,int
• teams.csv : int,str
• scores_full.csv : str,str,int,int,int,date,str,time,str,int

where date is formatted as a dd-mm-yyyy string and time a hh:mm string.

Usage

Depending on your workflow, this step may vary.

Using Rscript, you may run one of scoremodel-cont-var.R, scoremodel-disc-var.R, scoremodel-time-independent.R.

The model must currently be changed manually, by setting the MODEL variable. It can be [cont_var|cont_rw] in -cont-var.R or [disc_var|disc_rw|disc_wn] in -disc-var.R.

If you want to save the resulting parameters to file, you may use CLI arguments:

• -s --save [Ø|filename]
• -f, --format [json|csv|dat]

So saving the parameters for disc_var, you may run

Rscript scoremodel-disc-var.R --save --format json

The format should be specified when saving.

Notes

there are none.