"""
##############################################################################
# Import librairies
# -------------------------------------------
from museotoolbox.cross_validation import SpatialLeaveAsideOut
from museotoolbox import datasets, raster_tools, vector_tools
##############################################################################
# Load HistoricalMap dataset
# -------------------------------------------
raster, vector = datasets.historicalMap(low_res=True)
field = 'Class'
X, y = raster_tools.getSamplesFromROI(raster, vector, field)
distanceMatrix = vector_tools.getDistanceMatrix(raster, vector)
##############################################################################
# Create CV
# -------------------------------------------
# n_splits will be the number of the least populated class
SLOPO = SpatialLeaveAsideOut(valid_size=1 / 3,
n_splits=2,
distanceMatrix=distanceMatrix,
random_state=2)
print(SLOPO.get_n_splits(X, y))
###############################################################################
############################################################################## # Import librairies # ------------------------------------------- from museotoolbox.cross_validation import LeaveOneSubGroupOut from museotoolbox.raster_tools import getSamplesFromROI from museotoolbox import datasets ############################################################################## # Load HistoricalMap dataset # ------------------------------------------- raster,vector = datasets.historicalMap() field = 'Class' group = 'uniquefid' X,y,s = getSamplesFromROI(raster,vector,field,group) ############################################################################## # Create CV # ------------------------------------------- # if n_splits is False (default), the number of splits will be the smallest # number of subgroup of all labels. valid_size = 0.5 # Means 50% LOSGO = LeaveOneSubGroupOut(verbose=False,random_state=12) # ############################################################################### # .. note:: # Split is made to generate each fold LOSGO.get_n_splits(X,y,s)
from museotoolbox.cross_validation import SpatialLeaveOneSubGroupOut
from museotoolbox import datasets, raster_tools, vector_tools
##############################################################################
# Load HistoricalMap dataset
# -------------------------------------------
raster, vector, centroid = datasets.historicalMap(low_res=True, centroid=True)
field = 'Class'
##############################################################################
# Extract label ('Class' field) and groups ('uniquefid' field)
# Compute distanceMatrix with centroid (one point per group)
X, y, groups = raster_tools.getSamplesFromROI(raster, vector, field,
'uniquefid')
distanceMatrix, distanceLabel = vector_tools.getDistanceMatrix(
raster, centroid, 'uniquefid')
##############################################################################
# Create CV
# -------------------------------------------
# n_splits will be the number of the least populated class
SLOSGO = SpatialLeaveOneSubGroupOut(distanceThresold=100,
distanceMatrix=distanceMatrix,
distanceLabel=distanceLabel,
random_state=12)
###############################################################################
# .. note::
field = 'Class'
group = 'uniquefid'
##############################################################################
# Create CV
# -------------------------------------------
valid_size = 0.5 # Means 50%
LPSGO = LeavePSubGroupOut(valid_size=valid_size,
random_state=12,
verbose=False)
###############################################################################
# Extract X,y and group.
# -------------------------------------------
X, y, g = raster_tools.getSamplesFromROI(raster, vector, field, group)
###############################################################################
# .. note::
# Split is made to generate each fold
for tr, vl in LPSGO.split(X, y, g):
print(tr.shape, vl.shape)
print('y label with number of samples')
print(np.unique(y[tr], return_counts=True))
##############################################################################
# Differences with scikit-learn
# -------------------------------------------
from sklearn.model_selection import LeavePGroupsOut
# You need to specify the number of groups
# sklearn will compute different metrics, but will keep best results from kappa (refit='kappa')
LAP = learnAndPredict(n_jobs=1,verbose=1)
##############################################################################
# Create or use custom function
def reduceBands(X,bandToKeep=[0,2]):
# this function get the first and the last band
X=X[:,bandToKeep].reshape(-1,len(bandToKeep))
return X
# add this function to learnAndPredict class
LAP.customizeX(reduceBands)
# if you learn from vector, refit according to the f1_mean
X,y = getSamplesFromROI(raster,vector,field)
LAP.learnFromVector(X,y,cv=2,classifier=classifier,param_grid=dict(n_estimators=[10]),
scoring=scoring,refit='f1_mean')
# if you learn from raster
LAP.learnFromRaster(raster,vector,field,cv=2,classifier=classifier,param_grid=dict(n_estimators=[10]),
scoring=scoring,refit='f1_mean')
##############################################################################
# Read the model
# -------------------
print(LAP.model)
print(LAP.model.cv_results_)
print(LAP.model.best_score_)