# Model settings. Default values can be used unless users wwant to investigate various methods or formats.
# See ./docs/How_to_create_config_files.md for instructions

# master seed controlling all probabilistic process
master_seed = 20230104

########################################################################################################################
# local/global regression methods

########################################################################################################################
gridcore_continuous = 'LWR:Linear'
gridcore_classification = 'LWR:Logistic' # for probability of event
n_splits = 10 # only useful for sklearn machine learning methods

########################################################################################################################
# distance-based weight calculation
########################################################################################################################

# Number of stations to consider for each target point. nearstn_min<=nearstn_max.
nearstn_min = 35  # nearby stations: minimum number
nearstn_max = 35  # nearby stations: maximum number

# Weight calculation formula. See ./docs/How_to_create_config_files.md for instruction.
weight_formula = '(1 - (dist / maxdist) ** 3) ** 3'

# Initial Search Distance in km
initial_distance = 100

# first try this radius (km). if not enough, expand. Could be useful to reduce computation time for large domain search.
try_radius = 10000

########################################################################################################################
# default settings
########################################################################################################################
[netcdf_info]
# the spatial dims of input stations
stn_lat_name = 'lat'
stn_lon_name = 'lon'

# the 2D spatial dims of the target grid domain
# note that target grid domain netcdf must have x/y dims, while lat and lon are 2D arrays
grid_lat_name = 'latitude'
grid_lon_name = 'longitude'
grid_mask_name = 'mask'

# the 2D spatial dims of dynamic predictor inputs
dynamic_grid_lat_name = 'lat'
dynamic_grid_lon_name = 'lon'


[transform]
# note: the name must be consistent with transform_vars
[transform.boxcox]
exponent = 4


[sklearn]
# if no parameters are provided or if the section does not even exist, default parameters will be used.
[sklearn.RandomForestRegressor]
n_estimators = 500 # a example of RandomForestRegressor parameter
n_jobs = 5
[sklearn.RandomForestClassifier]
n_estimators = 500 # a example of RandomForestRegressor parameter
n_jobs = 5

[flags]
# output random fields
output_randomfield = false

# output filename format
append_date_to_output_filename = true

# print configuration information
print_config = false

# overwrite existing files
overwrite_merged_stnfile = false
overwrite_station_cc = false
overwrite_weight = false
overwrite_cv_reg = false
overwrite_grid_reg = false
overwrite_ens = false
overwrite_spcorr = false