fit_subintervalls: fit_subintervalls

In maxikellerbauer/stAirPol: Spatio-temporal Modelling of Air Pollution Data

Description

Fit the spatio-temporal models on subinterval.

Usage

fit_subintervalls(data, grid = NULL, training_set = NULL, unit = "1 week",
  nItr = 5000, nBurn = 1000, return = "prediction", tol = 0.01,
  retry_count = 5, retry_reason_mean = 500, retry_reason_rmse = 2000,
  mc.cores = parallel::detectCores() - 1, ...)

Arguments

data	Modelling data.frame which contains information about the covariables and the target variable
grid	grid data where a prediction should be calculated or a training_set
training_set	an object generated by get_test_and_training_set, if those an object is delivered, the data argument will be reduced to the in training_set specified training sensors.
unit	for splitting the dataset, see lubridate::floor_date for more informations
nItr	Number of MCMC iterations. Default value is 5000.
nBurn	Number of burn-in samples. This number of samples will be discarded before making any inference. Default value is 1000.
return	default is 'prediction', otherwise the parametervalues of each submodel is returned
tol	Minimum separation distance between any two locations out of those specified by coords, knots.coords and pred.coords. The default is 0.005. The programm will exit if the minimum distance is less than the non-zero specified value. This will ensure non-singularity of the covariance matrices.
retry_count	How much retries should be tried, if some strange behavior appears, for example posteriori Mean above 500 or a RMSE of the prediction above 2000. Both could be specified separately. Default is 5.
retry_reason_mean	retry on posteriori Mean above that value, default is 500
retry_reason_rmse	retry on a RMSE above that value, default is 2000
mc.cores	how much cores should be used for parallelization, default is one core less your maximum number of detected cores.
...	Parameters passed to fit_sp_model

Value

the grid with added columns of the prediction calculations, or the crossvaldiation predictions

See Also

fit_subintervalls, predict.stAirPol.model, predict_split fit_model

Examples

data("mini_dataset")
mini_dataset <- clean_model_data(mini_dataset)
formula = value ~ humi + temp + rainhist + windhist +
  trafficvol + log(sensor_age)
training_set <- get_test_and_training_set(mini_dataset, sampel_size = 0.75,
                                          random.seed = 220292)
prediction.gp <- fit_subintervalls(data = mini_dataset, formula = formula,
                            model = 'GP', training_set = training_set,
                            unit = '1 day')
retry model fitting and prediction in cause of a strange behavior
sometimes
remove sensors which are always NA

maxikellerbauer/stAirPol documentation built on May 3, 2019, 3:16 p.m.