Forecast file format

Forecasts will be held in csv files in a common format. Any model must produce forecasts in this format, and any comparison script should expect to see them in this format. Forecasts must be saved to a file in tmp that ends in [modelname]forecasts.csv. modelname can be anything, but forecasts.csv must be exact. It must also generate a list of AIC values, also with specific column names, saved to a file in tmp named [modelname]forecasts_model_aic.csv.

Columns of the forecasts.csv file are as follows:

date: date on which the model was run. In the format 'YYYY-MM-DD'. (If hindcasting is done then the date will be a later date than forecastyear or forecastmonth.)

forecastmonth: the month of a prediction (will be depreciated in the future)

forecastyear: the year of a prediction (will be depreciated in the future)

newmoonnumber: the new moon ID of the prediction (new moon 1 is the first sample in 1977)

currency: the type of value being predicted. Valid values are: c('abundance','richness','biomass','energy')

model: the name of the model

level: the spatial level of the site being predicted. Valid values are: c('All','Controls','FullExclosure','KratExclosure','Plot1','Plot2',...,'Plot24')

species: the species being predicted. Value are all the rodent species codes (c('DM','DL','OM'..etc)) or 'total' for the combined prediction of all species.

estimate: the mean value of the prediction

LowerPI: the lower bound of the predicted 90% CI

UpperPI: the upper bound of the predicted 90% CI

fit_start_newmoon: the first newmoonnumber in the data used to make the prediction

fit_end_newmoon: the last newmoonnumber in the data used to make the prediction

initial_newmoon: the first newmoonnumber for which predictions are made in the set

Note that currency, level, and species are the 3 unique things which models will be compared with. ie. a model predicting level: all, currency: abundance, species: all will only be compared against models making those same predictions.

Columns of the forecasts_model_aic.csv file are as follows:

date: date on which the model was run. In the format 'YYYY-MM-DD'. (If hindcasting is done then the date will be a later date than forecastyear or forecastmonth.)

currency: the type of value being predicted. Valid values are: c('abundance','richness','biomass','energy')

model: the name of the model

level: the spatial level of the site being predicted. Valid values are: c('All','Controls','FullExclosure','KratExclosure','Plot1','Plot2',...,'Plot24')

species: the species being predicted. Value are all the rodent species codes (c('DM','DL','OM'..etc)) or 'total' for the combined prediction of all species.

aic: AIC value

fit_start_newmoon: the first newmoonnumber in the data used to make the prediction

fit_end_newmoon: the last newmoonnumber in the data used to make the prediction

initial_newmoon: the first newmoonnumber for which predictions are made in the set