Since we use third-party datasets with different licenses/copyrights, we only include information on how we downloaded and processed the data.
Data description: https://fivethirtyeight.com/features/how-our-mlb-predictions-work/
Download this .csv file
inside this folder as data/mlb_elo_538.csv.
Then, load this dataset using
cc.data_utils.baseball.read_538_projections.
See also: nb_comparecast_baseball.ipynb.
Download these .xlsx files
inside this folder as data/mlb odds 20**.xlsx.
Then, load this dataset using
cc.data_utils.baseball.read_vegas_odds.
See also: nb_comparecast_baseball.ipynb.
Found in: eprob/replication_material/precip_fcs/*.csv.
These are downloaded from the eprob repository
and processed into .csv using R.
To re-run the R code, clone the eprob repository somewhere and
copy the content inside replication_material into $eprob_clone/replication_material.
Load the dataset using
cc.data_utils.weather.read_precip_fcs
(data & forecasts per location/lag)
or
cc.data_utils.weather.read_hz_evalues
(data & e-values only).
See also: nb_comparecast_weather.ipynb.