def test_variogram():
"""Test the return types"""
# get data
df = data.pancake()
hydrobox.plotting_backend('plotly')
for t, type_ in zip(('object', 'describe', 'plot'), (skg.Variogram, dict, go.Figure)):
vario = hydrobox.geostat.variogram(
df[['x', 'y']].values,
df.z.values,
return_type=t
)
assert isinstance(vario, type_)
def test_gridsearch():
"""Test Gridsearch with explicit param_grid"""
# create the grid
param_grid={'model': ('spherical', 'gaussian', 'stable')}
# get data
df = data.pancake()
gs = hydrobox.geostat.gridsearch(
param_grid=param_grid,
coordinates=df[['x', 'y']].values,
values=df.z.values,
n_lags=25,
return_type='object'
)
assert isinstance(gs, GridSearchCV)
def test_gridsearch_variogram():
"""Test Gridsearch with variogram passed"""
df = data.pancake()
# create the grid
param_grid={'fit_method': ('trf', 'lm', 'ml')}
vario = hydrobox.geostat.variogram(
df[['x', 'y']].values,
df.z.values,
n_lags=15,
return_type='object'
)
gs = hydrobox.geostat.gridsearch(
param_grid,
variogram=vario,
return_type='best_param'
)
assert isinstance(gs, dict)
plots. These plots help to understand the spatial properties of a variogram,
and finally, the :class:`Variogram <skgstat.Variogram>` object itself can be
returned and used in one of the Kriging routines.
"""
from pprint import pprint
import plotly
import hydrobox
from hydrobox.data import pancake
from hydrobox.plotting import plotting_backend
plotting_backend('plotly')
# %%
# Load sample data from the data sub-module
df = pancake()
# %%
# Estimate a variogram using a exponential model and 25 distance lags
# that are derived from a KMeans cluster algorithm
# Here, we use the describe output option to get a dictionary of
# all variogram parameters
vario = hydrobox.geostat.variogram(coordinates=df[['x', 'y']].values,
values=df.z.values,
model='exponential',
bin_func='kmeans',
n_lags=25,
return_type='describe')
# print
pprint(vario)