def test_TimeSeriesForest_predictions(n_estimators, n_intervals):
random_state = 1234
X_train, y_train = load_gunpoint(split="train", return_X_y=True)
X_test, y_test = load_gunpoint(split="test", return_X_y=True)
features = [np.mean, np.std, time_series_slope]
steps = [
(
"transform",
RandomIntervalFeatureExtractor(
random_state=random_state, features=features
),
),
("clf", DecisionTreeClassifier()),
]
estimator = Pipeline(steps)
clf1 = TimeSeriesForestClassifier(
estimator=estimator, random_state=random_state, n_estimators=n_estimators
)
clf1.fit(X_train, y_train)
a = clf1.predict_proba(X_test)
# default, semi-modular implementation using
# RandomIntervalFeatureExtractor internally
clf2 = TimeSeriesForestClassifier(
random_state=random_state, n_estimators=n_estimators
)
clf2.fit(X_train, y_train)
b = clf2.predict_proba(X_test)
np.testing.assert_array_equal(a, b)
def tsf_classifier(X_train, X_test, y_train, y_test):
"""ML Scorer based on sktime pipeline with a TimeSeriesForestClassifier."""
steps = [('concatenate', ColumnConcatenator()),
('classify', TimeSeriesForestClassifier(n_estimators=100))]
clf = Pipeline(steps)
clf.fit(X_train, y_train)
return clf.score(X_test, y_test)
def main():
generator = DataGenerator(labeled_data_file=args.labeled_data_file, data_util_file=args.data_util_file,
threshold=args.threshold, dt=args.dt, L=args.L, tmin=args.tmin, tmax=args.tmax)
training_data, test_data = generator.get_data(ts_nth_element=args.ts_nth_element,
training_frac=0.7)
steps = [
('extract', RandomIntervalFeatureExtractor(n_intervals='sqrt',
features=[np.mean, np.std, time_series_slope])),
('clf', DecisionTreeClassifier())
]
time_series_tree = Pipeline(steps)
tsf = TimeSeriesForestClassifier(
estimator=time_series_tree,
n_estimators=args.n_estimators,
criterion='entropy' if args.criterion == 'entropy' else 'gini',
bootstrap=True,
oob_score=True,
random_state=1,
# n_jobs=4,
verbose=1
)
x = detabularize(pd.DataFrame(training_data[:,1:]))
try:
with parallel_backend('threading', n_jobs=args.n_jobs):
tsf = tsf.fit(x, training_data[:,0])
with open('{save_file_name}.pickle'.format(save_file_name=args.save_file_name), 'wb') \
as TimeSeriesForestModel:
pickle.dump(tsf, TimeSeriesForestModel, protocol=pickle.HIGHEST_PROTOCOL)
except Exception as ex:
print(ex)
def rise_benchmarking():
for i in range(len(benchmark_datasets)):
dataset = benchmark_datasets[i]
print(str(i) + " problem = " + dataset)
rise = fb.RandomIntervalSpectralForest(n_estimators=100)
exp.run_experiment(overwrite=True,
problem_path=data_dir,
results_path=results_dir,
cls_name="PythonRISE",
classifier=rise,
dataset=dataset,
train_file=False)
steps = [('segment',
RandomIntervalSegmenter(n_intervals=1, min_length=5)),
('transform',
FeatureUnion([('acf',
RowTransformer(
FunctionTransformer(func=acf_coefs,
validate=False))),
('ps',
RowTransformer(
FunctionTransformer(func=powerspectrum,
validate=False)))])),
('tabularise', Tabularizer()),
('clf', DecisionTreeClassifier())]
base_estimator = Pipeline(steps)
rise = TimeSeriesForestClassifier(estimator=base_estimator,
n_estimators=100)
exp.run_experiment(overwrite=True,
problem_path=data_dir,
results_path=results_dir,
cls_name="PythonRISEComposite",
classifier=rise,
dataset=dataset,
train_file=False)
def _rf_scorer(X_train, X_test, y_train, y_test):
steps = [
('concatenate', ColumnConcatenator()),
('classify', TimeSeriesForestClassifier(n_estimators=100))
]
clf = Pipeline(steps)
clf.fit(X_train, y_train)
return clf.score(X_test, y_test)
def tsf_benchmarking():
for i in range(0, len(benchmark_datasets)):
dataset = benchmark_datasets[i]
print(str(i) + " problem = " + dataset)
tsf = ib.TimeSeriesForest(n_estimators=100)
exp.run_experiment(
overwrite=False,
problem_path=data_dir,
results_path=results_dir,
cls_name="PythonTSF",
classifier=tsf,
dataset=dataset,
train_file=False,
)
steps = [
("segment", RandomIntervalSegmenter(n_intervals="sqrt")),
(
"transform",
FeatureUnion(
[
(
"mean",
make_row_transformer(
FunctionTransformer(func=np.mean, validate=False)
),
),
(
"std",
make_row_transformer(
FunctionTransformer(func=np.std, validate=False)
),
),
(
"slope",
make_row_transformer(
FunctionTransformer(
func=time_series_slope, validate=False
)
),
),
]
),
),
("clf", DecisionTreeClassifier()),
]
base_estimator = Pipeline(steps)
tsf = TimeSeriesForestClassifier(estimator=base_estimator, n_estimators=100)
exp.run_experiment(
overwrite=False,
problem_path=data_dir,
results_path=results_dir,
cls_name="PythonTSFComposite",
classifier=tsf,
dataset=dataset,
train_file=False,
)
def test_predict_proba():
clf = TimeSeriesForestClassifier(n_estimators=2)
clf.fit(X, y)
proba = clf.predict_proba(X)
assert proba.shape == (X.shape[0], n_classes)
np.testing.assert_array_equal(np.ones(X.shape[0]), np.sum(proba, axis=1))
# test single row input
y_proba = clf.predict_proba(X.iloc[[0], :])
assert y_proba.shape == (1, n_classes)
y_pred = clf.predict(X.iloc[[0], :])
assert y_pred.shape == (1,)
def fit(self, luck_average_windows, assessment_windows, until=None, max_horizon=9 * 6):
logger("MODEL-FIT").debug(
"max_horizon: {} / avg windows: {} / assmnt windows: {} / until: {} / total_data_size: {}".format(
max_horizon,
str(luck_average_windows),
str(assessment_windows),
until,
len(self.data_points)))
if until is not None and (until < 0 or until >= len(self.data_points)):
logger("MODEL-FIT").error("Parameter until is too large for the given data points: {}".format(until))
return
self.horizon = max_horizon
for wi, w in enumerate(assessment_windows):
if w > self.horizon:
break
# prepare data frame for sktime package
temporary_data_fit_file = self.prepare_ts_file(0, len(self.data_points) if until is None else until,
self.case_observation_size, wi, w)
# parse data frames from the temporary fit data file
X, y = load_from_tsfile_to_dataframe(temporary_data_fit_file, replace_missing_vals_with="-100")
# which label is the first one?
true_index = 0
if y[0] == "false":
true_index = 1
new_class_weights = self.create_class_weight_dict(true_index=true_index)
estimators = []
for i in range(0, len(luck_average_windows)):
estimators.append(("TSF{}".format(i), TimeSeriesForestClassifier(
n_estimators=int(self.no_estimators),
n_jobs=16,
max_depth=self.max_depth,
class_weight=new_class_weights,
criterion=self.criterion,
min_samples_split=self.min_samples_split,
min_samples_leaf=self.min_samples_leaf,
oob_score=self.oob_score,
bootstrap=self.bootstrap),
[i]))
c = ColumnEnsembleClassifier(estimators=estimators)
c.fit(X, y)
# print(str(c.classes_))
self.classifiers.append(c)
def main(args):
# Load and wrangle data
raw_data_df = run.input_datasets["rawdata"].to_pandas_dataframe()
processed_data_df = prepare_dataframe(
raw_data_df,
time_series_length=args.timeserieslength,
threshold=args.threshold)
# Split data
train = processed_data_df.sample(frac=args.train_data_split,
random_state=42)
test = processed_data_df.drop(train.index)
# Example for logging
run.log(
"data_split_fraction",
args.train_data_split,
"Fraction of samples used for training",
)
run.log("train_samples", train.shape[0],
"Number of samples used for training")
run.log("test_samples", test.shape[0],
"Number of samples used for testing")
# Train
task = TSCTask(target="label", metadata=train)
clf = TimeSeriesForestClassifier(n_estimators=args.n_estimators)
strategy = TSCStrategy(clf)
strategy.fit(task, train)
run.log("n_estimators", args.n_estimators,
"Number of tree estimators used in the model")
# Metrics
y_pred = strategy.predict(test)
y_test = test[task.target]
accuracy = accuracy_score(y_test, y_pred)
run.log("Accuracy", f"{accuracy:1.3f}", "Accuracy of model")
# Persist model
os.makedirs("outputs", exist_ok=True)
model_path = os.path.join("outputs", args.model_filename)
dump(strategy, model_path)
def test_stat():
data = load_gunpoint(split="train")
dataset = RAMDataset(dataset=data, name="gunpoint")
task = TSCTask(target="class_val")
fc = TimeSeriesForestClassifier(n_estimators=1, random_state=1)
strategy_fc = TSCStrategy(fc, name="tsf")
pf = ProximityForest(n_estimators=1, random_state=1)
strategy_pf = TSCStrategy(pf, name="pf")
# result backend
results = RAMResults()
orchestrator = Orchestrator(
datasets=[dataset],
tasks=[task],
strategies=[strategy_pf, strategy_fc],
cv=SingleSplit(random_state=1),
results=results,
)
orchestrator.fit_predict(save_fitted_strategies=False)
analyse = Evaluator(results)
metric = PairwiseMetric(func=accuracy_score, name="accuracy")
_ = analyse.evaluate(metric=metric)
ranks = analyse.rank(ascending=True)
pf_rank = ranks.loc[ranks.strategy == "pf",
"accuracy_mean_rank"].item() # 1
fc_rank = ranks.loc[ranks.strategy == "tsf",
"accuracy_mean_rank"].item() # 2
rank_array = [pf_rank, fc_rank]
rank_array_test = [1, 2]
_, sign_test_df = analyse.sign_test()
sign_array = [
[sign_test_df["pf"][0], sign_test_df["pf"][1]],
[sign_test_df["tsf"][0], sign_test_df["tsf"][1]],
]
sign_array_test = [[1, 1], [1, 1]]
np.testing.assert_equal([rank_array, sign_array],
[rank_array_test, sign_array_test])
def tsf_benchmarking():
for i in range(len(benchmark_datasets)):
dataset = benchmark_datasets[i]
print(str(i) + " problem = " + dataset)
tsf = ib.TimeSeriesForest(n_estimators=100)
exp.run_experiment(overwrite=False,
problem_path=data_dir,
results_path=results_dir,
cls_name="PythonTSF",
classifier=tsf,
dataset=dataset,
train_file=False)
steps = [
('segment', RandomIntervalSegmenter(n_intervals='sqrt')),
('transform',
FeatureUnion([('mean',
RowTransformer(
FunctionTransformer(func=np.mean,
validate=False))),
('std',
RowTransformer(
FunctionTransformer(func=np.std,
validate=False))),
('slope',
RowTransformer(
FunctionTransformer(func=time_series_slope,
validate=False)))])),
('clf', DecisionTreeClassifier())
]
base_estimator = Pipeline(steps)
tsf = TimeSeriesForestClassifier(estimator=base_estimator,
n_estimators=100)
exp.run_experiment(overwrite=False,
problem_path=data_dir,
results_path=results_dir,
cls_name="PythonTSFComposite",
classifier=tsf,
dataset=dataset,
train_file=False)
def main():
#1. Loading and splitting the dataset
X_train, y_train = load_italy_power_demand(split='train', return_X_y=True)
X_test, y_test = load_italy_power_demand(split='test', return_X_y=True)
print('Shape of X, y train and test dataset', X_train.shape, y_train.shape,
X_test.shape, y_test.shape, '\n')
print('X_train:', X_train.head(), '\n')
print('\nX_train info', X_train.info(), '\n')
labels, counts = np.unique(y_train, return_counts=True)
print(
'\nThere are', labels,
'labels in this dataset, one corresponds to winter and the other to summer. The counter of each one is',
counts, '\n')
#2. Creating a Model, Fit and Predict Sklearn Classifier
#Sktime Tabularizing the data
X_train_tab = tabularize(X_train)
X_test_tab = tabularize(X_test)
print('\n X_train tabularized\n', X_train_tab.head(), '\n')
#2.1 SKlearn RandomForest Classifier
classifier = RandomForestClassifier(n_estimators=100)
classifier.fit(X_train_tab, y_train)
y_pred = classifier.predict(X_test_tab)
print('Accuracy sklearn RandomForestClassifier',
round(accuracy_score(y_test, y_pred), 4), '\n')
#2.2 Same SKlearn as above but using make_pipeline w/ Sktime Tabularizer
classifier = make_pipeline(Tabularizer(),
RandomForestClassifier(n_estimators=100),
verbose=True)
classifier.fit(X_train, y_train)
print(
'Accuracy sklearn RandomForestClassifier using sklearn make_pipeline in which the first step is to sktime Tabularize()',
round(classifier.score(X_test, y_test), 4), '\n')
#3 Sklearn using make_pipeline w/ Sktime TSFreshFeatureExtractor
classifier = make_pipeline(TSFreshFeatureExtractor(show_warnings=False),
RandomForestClassifier(n_estimators=100))
classifier.fit(X_train, y_train)
print(
'Accuracy sklearn RandomForestClassifier using sklearn make_pipeline in which the first step is to sktime TSFreshFeatureExtractor that automatically extracts and filters several key statistical features from the nested X_train time series',
round(classifier.score(X_test, y_test), 4), '\n')
#4. Using Time series algorithms and classifiers from sklearn/sktime
steps = [
('segment', RandomIntervalSegmenter(n_intervals='sqrt')), #Sktime
(
'transform',
FeatureUnion([ #Sklearn
('mean',
RowTransformer(
FunctionTransformer(func=np.mean,
validate=False))), #sktime
('std',
RowTransformer(
FunctionTransformer(func=np.std,
validate=False))), #sktime
('slope',
RowTransformer(
FunctionTransformer(func=time_series_slope,
validate=False))) #sktime
])),
('clf', DecisionTreeClassifier()) #From Sklearn
]
time_series_tree = Pipeline(steps, verbose=True) #sklearn
time_series_tree.fit(X_train, y_train)
print(
'Accuracy sklearn DecisionTreeClassifier using sklearn Pipeline() as well as segmentation and transformation techniques from sktime and sklearn',
round(time_series_tree.score(X_test, y_test), 4))
#5. Using Time series Sktime
tsf = TimeSeriesForestClassifier(n_estimators=100, verbose=True)
tsf.fit(X_train, y_train)
print('Accuracy sktime TimeSeriesForestClassifier',
round(tsf.score(X_test, y_test), 4))
# data -> our function -> (X_nested, y)
X = generate_long_table(ts)
X.head()
X_nested = from_long_to_nested(X)
X_nested.head()
y = np.array(['a']) # , 'b', 'a', 'b', 'a', 'b', 'a', 'b'])
print(X_nested)
X_train, X_test, y_train, y_test = train_test_split(X_nested, y)
print(X.head())
classifier = ColumnEnsembleClassifier(estimators=[
("TSF1", TimeSeriesForestClassifier(n_estimators=100), [1]),
("TSF2", TimeSeriesForestClassifier(n_estimators=100), [2]),
])
classifier.fit(X_train, y_train)
# Use the test portion of data for prediction so we can understand how accurate our model was learned
y_pred = classifier.predict(X_test)
# Use the native `accuracy_score` method to calculate the accuracy based on the test outcomes and the predicted outcomes
print("Accuracy score is: " + str(accuracy_score(y_test, y_pred)))
def generate_example_long_table(num_cases=50, series_len=20, num_dims=2):
rows_per_case = series_len * num_dims
total_rows = num_cases * series_len * num_dims
case_ids = np.empty(total_rows, dtype=np.int)
X_train_timedata = X_train_timedata.to_frame()
X_test_timedata = X_test_timedata.to_frame()
ts_train = pd.Series(X_train_timedata['combine'].values,
index=X_train_timedata.index)
X_ts_train = ts_train.to_frame()
ts_test = pd.Series(X_test_timedata['combine'].values,
index=X_test_timedata.index)
X_ts_test = ts_test.to_frame()
for row_num in range(0, X_ts_train.shape[0]):
series1 = pd.Series(X_ts_train.iat[row_num, 0])
X_ts_train.iat[row_num, 0] = series1
for row_num in range(0, X_ts_test.shape[0]):
series2 = pd.Series(X_ts_test.iat[row_num, 0])
X_ts_test.iat[row_num, 0] = series2
## =======================Column ensembling================================
clf = ColumnEnsembleClassifier(estimators=[
("TSF0", TimeSeriesForestClassifier(n_estimators=5), [0]),
])
start_time = time.time()
clf.fit(X_ts_train, y_train)
Efficiency = time.time() - start_time
Accuracy = clf.score(X_ts_test, y_test)
print("Efficiency is:\n", Efficiency)
print("Accuracy is :\n", Accuracy)
signal_names = ["chan_%d" % x for x in range(num_channels)]
return signal_names, X, y
def testlime(signal_names, clf, x, y):
class_names=[y]
num_slices=20
num_features=10
explainer = lime_ts.LimeTimeSeriesExplainer(class_names=class_names,
signal_names=signal_names)
labelid = 0
exp = explainer.explain_instance(x, clf.predict_proba, num_features=num_features, num_samples=100, num_slices=num_slices, labels=[labelid], replacement_method='total_mean')
exp.as_pyplot_figure(labelid)
plt.show()
if __name__ == '__main__':
logging.getLogger().setLevel(logging.INFO)
signal_names, X, y = genDataset(80, 4, 30)
steps = [
("concatenate", ColumnConcatenator()),
("classify", TimeSeriesForestClassifier(n_estimators=100)),
]
clf = Pipeline(steps)
clf.fit(X,y)
testlime(signal_names, clf, X[0], y[0])
models = {
"features":
make_pipeline(TruncationTransformer(lower=MAX_LENGTH),
TSFreshFeatureExtractor(default_fc_parameters="efficient",
show_warnings=False,
n_jobs=-1),
RandomForestClassifier(n_jobs=-1, random_state=1),
verbose=True),
"interval":
make_pipeline(TruncationTransformer(lower=15000),
TimeSeriesForestClassifier(
estimator=time_series_tree,
n_estimators=100,
criterion="entropy",
bootstrap=True,
oob_score=True,
random_state=1,
n_jobs=-1,
),
verbose=True),
"shapelet":
make_pipeline(TruncationTransformer(lower=1000),
ContractedShapeletTransform(
time_contract_in_mins=10,
num_candidates_to_sample_per_case=10,
verbose=2,
random_state=1),
RandomForestClassifier(n_estimators=100,
n_jobs=-1,
random_state=1),
random_state=j)
# set CV
_, counts = np.unique(y_train, return_counts=True)
n_splits = np.minimum(counts.min(), INNER_N_SPLITS)
n_repeats = np.maximum(1, INNER_N_SPLITS // n_splits)
# cv = StratifiedKFold(n_splits=n_splits, shuffle=True,
# random_state=RANDOM_STATE)
inner_cv = RepeatedStratifiedKFold(n_splits=n_splits,
n_repeats=n_repeats,
random_state=RANDOM_STATE)
print(f'Dataset: {i + 1}/{n_datasets} {dataset.name} - n_splits: '
f'{j + 1}/{OUTER_CV_N_SPLITS}')
# set estimator
estimator = TimeSeriesForestClassifier(BASE_ESTIMATOR, n_jobs=-1)
gscv = GridSearchCV(estimator,
param_grid,
scoring='neg_log_loss',
cv=inner_cv,
refit=True,
iid=False,
error_score='raise',
verbose=True)
# tune when enough samples for all classes are available
start = time.time()
gscv.fit(X_train, y_train)
results[0] = time.time() - start
# predict
X_test = tslearn.utils.to_sktime_dataset(X_test)
y_train = np.vstack([y_train, y_val])
y_train = pd.Series(y_train.reshape(-1))
y_test = pd.Series(y_test.reshape(-1))
# Timeseries random foreset for every column
for i, col in enumerate(col_names[:2]):
print(col)
# Choose one feature
X_train_step = X_train.iloc[:, [i]]
X_test_step = X_test.iloc[:, [i]]
# Time series forest clf
classifier = TimeSeriesForestClassifier()
classifier.fit(X_train_step, y_train)
y_pred = classifier.predict(X_test_step)
# Metrics
print(f'accuracy_test: {accuracy_score(y_test, y_pred)}')
print(f"recall_test: {recall_score(y_test, y_pred)}")
print(f"precisoin_test: {precision_score(y_test, y_pred)}")
print(f"f1_test: {f1_score(y_test, y_pred)}")
# clf2 = pickle.loads(s)
# clf2.predict(X_test[0:1])
def main(model, input_training_raster, train_feature, input_test_raster,
test_feature, input_test_csv, result_path, n_channels, n_jobs,
model_path, raster_to_classify, patch_size, output_raster,
train_ratio, n_estimators, max_depth, max_num_of_samples_per_class):
# -- Creating output path if does not exist
if not os.path.exists(result_path):
os.makedirs(result_path)
# ---- output files
result_path = os.path.join(result_path, model)
if not os.path.exists(result_path):
os.makedirs(result_path)
print("Model: ", model)
# Generatin train/test datasets
train_list, test_list, _ = split_train_feature(train_feature, train_ratio)
train_data = generate_training_data(input_training_raster, train_feature,
train_list,
max_num_of_samples_per_class)
X_train, y_train = train_data[:, 1:], train_data[:, 0]
if input_test_raster and test_feature:
_, test_list, _ = split_train_feature(test_feature, train_ratio=0)
test_data = generate_training_data(input_test_raster, test_feature,
test_list)
X_test, y_test = test_data[:, 1:], test_data[:, 0]
elif input_test_csv:
df = pd.read_csv(input_test_csv, sep=',', header=None)
test_data = np.asarray(df.values)
X_test, y_test = test_data[:, 2:], test_data[:, 0]
else:
test_data = generate_training_data(input_training_raster,
train_feature, test_list,
max_num_of_samples_per_class)
X_test, y_test = test_data[:, 1:], test_data[:, 0]
# Fitting the classifier into the Training set
n_classes_test = len(np.unique(y_test))
n_classes_train = len(np.unique(y_train))
if (n_classes_test != n_classes_train):
print("WARNING: different number of classes in train and test")
n_classes = max(n_classes_train, n_classes_test)
# Torch, numpy, whatever, all index from 0, if we did not assign landcover classes
# with [0, 1, 2, 3, ...], it may cause problem, things get easier by reindex classes
lc_ids_old = np.unique(y_train)
lc_ids_old.sort()
lc_ids_new = np.arange(n_classes_train)
indexes = [np.where(y_train == lc_id)[0] for lc_id in lc_ids_old]
for index, lc_id_new in zip(indexes, lc_ids_new):
y_train[index] = lc_id_new
indexes = [np.where(y_test == lc_id)[0] for lc_id in lc_ids_old]
for index, lc_id_new in zip(indexes, lc_ids_new):
y_test[index] = lc_id_new
relation = np.vstack((lc_ids_old, lc_ids_new))
if model in ["RF", "SVM"]:
is_ts = False
# ---- Normalizing the data per band,
min_per = np.percentile(X_train, 2, axis=(0))
max_per = np.percentile(X_train, 100 - 2, axis=(0))
X_train = (X_train - min_per) / (max_per - min_per)
X_test = (X_test - min_per) / (max_per - min_per)
if model == "RF":
clf = RandomForestClassifier(n_estimators=n_estimators,
max_depth=max_depth,
criterion='entropy',
random_state=None,
verbose=0,
n_jobs=n_jobs)
elif model == "SVM":
clf = OneVsRestClassifier(
BaggingClassifier(SVC(kernel='linear', cache_size=200),
max_samples=1.0,
n_estimators=n_estimators,
verbose=0,
n_jobs=n_jobs))
elif model == "RF_TS":
from sktime.classification.compose import TimeSeriesForestClassifier
from sktime.transformations.panel.compose import ColumnConcatenator
is_ts = True
X_train = X_train.reshape(X_train.shape[0],
int(X_train.shape[1] / n_channels),
n_channels)
X_test = X_test.reshape(X_test.shape[0],
int(X_test.shape[1] / n_channels), n_channels)
# ---- Normalizing the data per band,
min_per = np.percentile(X_train, 2, axis=(0, 1))
max_per = np.percentile(X_train, 100 - 2, axis=(0, 1))
X_train = (X_train - min_per) / (max_per - min_per)
X_test = (X_test - min_per) / (max_per - min_per)
steps = [
("concatenate", ColumnConcatenator()),
("classify",
TimeSeriesForestClassifier(n_estimators=n_estimators,
max_depth=max_depth,
n_jobs=n_jobs)),
]
clf = Pipeline(steps)
# Train classifier
clf.fit(X_train, y_train)
# Save trained classifier
if not model_path:
model_path = os.path.join(result_path, 'Best_model.pkl')
joblib.dump(clf, model_path)
# Evaluation
start = time.time()
y_pred = clf.predict(X_test)
Classes = [f'class {i}' for i in np.unique(y_test)]
scores = metrics(y_test, y_pred, Classes)
scores_msg = ", ".join([f"{k}={v}" for (k, v) in scores.items()])
scores["time"] = (time.time() - start) / 60
log = {k: [v] for k, v in scores.items()}
log_df = pd.DataFrame(log)
log_df.to_csv(os.path.join(result_path, "trainlog.csv"))
print(
scores["report"]
) # In report, precision means User_accuracy, recall means Producer_accuracy
print(scores["confusion_matrix"])
# ---- Save min_max
minMaxVal_file = os.path.join(result_path, 'min_Max.txt')
save_minMaxVal(minMaxVal_file, min_per, max_per)
# Inference on raster
if raster_to_classify:
classify_image(raster_to_classify,
model_path,
output_raster,
n_channels,
patch_size=patch_size,
minmax=[min_per, max_per],
is_ts=is_ts,
relation=relation)
import pytest
from sktime.benchmarking.strategies import TSCStrategy
from sktime.benchmarking.tasks import TSCTask
from sktime.datasets import load_gunpoint
from sktime.datasets import load_italy_power_demand
from sktime.classification.compose import TimeSeriesForestClassifier
classifier = TimeSeriesForestClassifier(n_estimators=2)
DATASET_LOADERS = (load_gunpoint, load_italy_power_demand)
# Test output of time-series classification strategies
@pytest.mark.parametrize("dataset", DATASET_LOADERS)
def test_TSCStrategy(dataset):
train = dataset(split='train')
test = dataset(split='test')
s = TSCStrategy(classifier)
task = TSCTask(target='class_val')
s.fit(task, train)
y_pred = s.predict(test)
assert y_pred.shape == test[task.target].shape