def test_impute():
from pyimpute import load_training_rasters, load_targets, impute
# Load training data
train_xs, train_y = load_training_rasters(response_raster,
explanatory_rasters)
# Train a classifier
from sklearn.ensemble import RandomForestClassifier
clf = RandomForestClassifier(n_estimators=10, n_jobs=1)
clf.fit(train_xs, train_y)
# Load targets
target_xs, raster_info = load_targets(explanatory_rasters)
# Go...
impute(target_xs,
clf,
raster_info,
outdir=TMPOUT,
linechunk=400,
class_prob=True,
certainty=True)
assert os.path.exists(os.path.join(TMPOUT, "responses.tif"))
assert os.path.exists(os.path.join(TMPOUT, "certainty.tif"))
assert os.path.exists(os.path.join(TMPOUT, "probability_90.tif"))
def main():
# Define the known data points or "training" data
explanatory_fields = "tmin12c tmax8c p_ph_c pmean_wntrc pmean_sumrc irr_lands gt_demc grwsnc d2u2c".split()
explanatory_rasters = [os.path.join(TRAINING_DIR, r, "hdr.adf") for r in explanatory_fields]
response_raster = os.path.join(TRAINING_DIR, 'iso_zns3-27/hdr.adf')
# Take a random stratified sample
selected = stratified_sample_raster(response_raster,
target_sample_size=20, min_sample_proportion=0.01)
# Load the training rasters using the sampled subset
train_xs, train_y = load_training_rasters(response_raster,
explanatory_rasters, selected)
print(train_xs.shape, train_y.shape)
# Train the classifier
clf = ExtraTreesClassifier(n_estimators=10, n_jobs=1)
clf.fit(train_xs, train_y)
print(clf)
# Cross validate
k = 5
scores = cross_validation.cross_val_score(clf, train_xs, train_y, cv=k)
print("%d-fold Cross Validation Accuracy: %0.2f (+/- %0.2f)" % (k, scores.mean() * 100, scores.std() * 200))
# ... Other model assessment
# Run the model on the current data; i.e. predict itself
print("Imputing response rasters FOR CURRENT DATA")
target_xs, raster_info = load_targets(explanatory_rasters)
impute(target_xs, clf, raster_info, outdir="_aez_output_current",
linechunk=400, class_prob=True, certainty=True)
sys.exit()
years = ['2070s']
for year in years:
print("Loading target explanatory raster data, swapping out for %s climate data" % year)
fdir = os.path.join(TRAINING_DIR, "../RCP85/%s/" % year)
# swap out datasets that are predicted to change over time (i.e the climate data only)
climate_rasters = "grwsnc pmean_sumrc pmean_wntrc tmax8c tmin12c".split()
new_explanatory_rasters = OrderedDict(zip(explanatory_fields, explanatory_rasters))
for cr in climate_rasters:
new_explanatory_rasters[cr] = fdir + cr + "/hdr.adf"
target_xs, raster_info = load_targets(new_explanatory_rasters.values())
print("Imputing response rasters")
impute(target_xs, clf, raster_info, outdir="_aez_output_%s" % year,
linechunk=40, class_prob=True, certainty=True)
def main():
# Define the known data points or "training" data
explanatory_fields = "d100 dd0 dd5 fday ffp gsdd5 gsp map mat_tenths mmax_tenths mmindd0 mmin_tenths mtcm_tenths mtwm_tenths sday".split()
explanatory_rasters = [os.path.join(TRAINING_DIR, "current_" + r + ".img") for r in explanatory_fields]
response_shapes = os.path.join(TRAINING_DIR, "DF.shp")
# Load the training rasters using the sampled subset
try:
cached = json.load(open("_cached_training.json"))
train_xs = np.array(cached['train_xs'])
train_y = np.array(cached['train_y'])
except IOError:
train_xs, train_y = load_training_vector(response_shapes,
explanatory_rasters, response_field='GRIDCODE')
cache = {'train_xs': train_xs.tolist(), 'train_y': train_y.tolist()}
with open("_cached_training.json", 'w') as fh:
fh.write(json.dumps(cache))
print(train_xs.shape, train_y.shape)
# Train the classifier
clf = ExtraTreesClassifier(n_estimators=120, n_jobs=3)
clf.fit(train_xs, train_y)
print(clf)
# Cross validate
k = 5
scores = cross_validation.cross_val_score(clf, train_xs, train_y, cv=k)
print("%d-fold Cross Validation Accuracy: %0.2f (+/- %0.2f)" % (k, scores.mean() * 100, scores.std() * 200))
# Run the model on the current data; i.e. predict current conditions
print("Imputing response rasters FOR CURRENT DATA")
target_xs, raster_info = load_targets(explanatory_rasters)
impute(target_xs, clf, raster_info, outdir="_usfs_output_current",
linechunk=400, class_prob=True, certainty=True)
years = ['2060']
for year in years:
print("Loading target explanatory raster data, swapping out for %s climate data" % year)
# Swap out for future climate rasters
new_explanatory_rasters = [os.path.join(TRAINING_DIR, "Ensemble_rcp60_y%s_%s.img" % (year, r))
for r in explanatory_fields]
target_xs, raster_info = load_targets(new_explanatory_rasters)
print("Imputing response rasters")
impute(target_xs, clf, raster_info, outdir="_usfs_output_%s" % year,
linechunk=400, class_prob=True, certainty=True)
def test_impute():
from pyimpute import load_training_rasters, load_targets, impute
# Load training data
train_xs, train_y = load_training_rasters(response_raster, explanatory_rasters)
# Train a classifier
from sklearn.ensemble import RandomForestClassifier
clf = RandomForestClassifier(n_estimators=10, n_jobs=1)
clf.fit(train_xs, train_y)
# Load targets
target_xs, raster_info = load_targets(explanatory_rasters)
# Go...
impute(target_xs, clf, raster_info, outdir=TMPOUT,
linechunk=400, class_prob=True, certainty=True)
assert os.path.exists(os.path.join(TMPOUT, "responses.tif"))
assert os.path.exists(os.path.join(TMPOUT, "certainty.tif"))
assert os.path.exists(os.path.join(TMPOUT, "probability_90.tif"))
def main():
# Define the known data points or "training" data
explanatory_fields = "d100 dd0 dd5 fday ffp gsdd5 gsp map mat_tenths mmax_tenths mmindd0 mmin_tenths mtcm_tenths mtwm_tenths sday".split(
)
explanatory_rasters = [
os.path.join(TRAINING_DIR, "current_" + r + ".img")
for r in explanatory_fields
]
response_shapes = os.path.join(TRAINING_DIR, "DF.shp")
# Load the training rasters using the sampled subset
try:
cached = json.load(open("_cached_training.json"))
train_xs = np.array(cached['train_xs'])
train_y = np.array(cached['train_y'])
except IOError:
train_xs, train_y = load_training_vector(response_shapes,
explanatory_rasters,
response_field='GRIDCODE')
cache = {'train_xs': train_xs.tolist(), 'train_y': train_y.tolist()}
with open("_cached_training.json", 'w') as fh:
fh.write(json.dumps(cache))
print(train_xs.shape, train_y.shape)
# Train the classifier
clf = ExtraTreesClassifier(n_estimators=120, n_jobs=3)
clf.fit(train_xs, train_y)
print(clf)
# Cross validate
k = 5
scores = cross_validation.cross_val_score(clf, train_xs, train_y, cv=k)
print("%d-fold Cross Validation Accuracy: %0.2f (+/- %0.2f)" %
(k, scores.mean() * 100, scores.std() * 200))
# Run the model on the current data; i.e. predict current conditions
print("Imputing response rasters FOR CURRENT DATA")
target_xs, raster_info = load_targets(explanatory_rasters)
impute(target_xs,
clf,
raster_info,
outdir="_usfs_output_current",
linechunk=400,
class_prob=True,
certainty=True)
years = ['2060']
for year in years:
print(
"Loading target explanatory raster data, swapping out for %s climate data"
% year)
# Swap out for future climate rasters
new_explanatory_rasters = [
os.path.join(TRAINING_DIR, "Ensemble_rcp60_y%s_%s.img" % (year, r))
for r in explanatory_fields
]
target_xs, raster_info = load_targets(new_explanatory_rasters)
print("Imputing response rasters")
impute(target_xs,
clf,
raster_info,
outdir="_usfs_output_%s" % year,
linechunk=400,
class_prob=True,
certainty=True)