from plantstat.data_generators import ClusterData
data_gen = ClusterData(n_features = 5, n_samples = 1000, cluster_std = 1.2,
                       centers_range = (4, 5), random_state = 0, return_labels = False)
X = pd.DataFrame(data_gen.generate(save = True))

from plantstat.data_generators import RegressionData
data_gen = RegressionData(n_features = 5, n_samples = 1000, n_informative = 3, n_targets = 1,
                          bias = 0.0, noise = 0.2, shuffle = True, random_state = 0, return_labels = True)
X, y = data_gen.generate(save = True)

from plantstat import Variable_Analyzer
# Define Analyzer
a = Variable_Analyzer(data, labels)

# An example of step-by-step analysis with saving in a local directory
# EDA
a.boxplot(save = True)
a.outliers()
a.corrs(method = 'pearson', heatmap = True, save = True)
a.QQplot(save = True)
a.pair_plot(save = True)

# Basic data statistics
a.basic_stats(p_value = False)
a.basic_stats(p_value = True, save = True)

# Statistical tests
a.var_compare(p_value = False)
a.var_compare(p_value = True, save = True)

from plantstat import AutoML_Classifier
# Define AutoML_Classifier
model = AutoML_Classifier(n_iter = 100)
model.fit(X_train, y_train)

# Model detailed information
# model.cv_results_
# model.best_estimator_
# model.best_pipeline

# Prediction and classification report (with prediction saving)
model.predict(X_test, save = True)
model.predict_proba(X_test, save = True, f_format = 'csv')
model.classification_report(X_test, y_test, labels = class_names, cmap = 'cividis', save = True)

# AutoML model without some algorithms
model = AutoML_Classifier(n_iter = 100, XGB = False, GradientBoosting = False)
model.fit(X_train, y_train)
model.classification_report(X_test, y_test, labels = class_names)

from plantstat import AutoML_Regressor
# Define AutoML_Classifier
model = AutoML_Regressor(n_iter = 100)
model.fit(X_train, y_train)

# Model detailed information
# model.cv_results_
# model.best_estimator_
# model.best_pipeline

# Predcition and report (with prediction saving)
model.predict(X_test, save = True)
model.prediction_report(X_test, y_test, save = True)

# AutoML model without some algorithms
model = AutoML_Regressor(n_iter = 100, XGB = False, GradientBoosting = False)
model.fit(X_train, y_train)
model.prediction_report(X_test, y_test)

from plantstat import Clusterer
from plantstat.data_generators import ClusterData

# Create syntetic data with 5 features, 1000 samples and 4 clusters
data_gen = ClusterData(n_features = 5, n_samples = 1000, cluster_std = 1.2,
                       centers_range = (4, 5), random_state = 0)
X = pd.DataFrame(data_gen.generate())

# Create K-means model for clustering (the model includes PCA with 2 components)
kmeans = Clusterer(is_pca = True, clusterer = 'kmeans')
kmeans.fit(X, save = True)

# Score values for various number of clusters
kmeans.scores

# Preprocessed data (after scaling and PCA)
kmeans.X

# Prediction (basis on fit results)
preds = kmeans.predict(k = 4, save = True)

# The real number of labels
print('Unique labels: %i' %len(np.unique(data_gen.labels)))


# Create DBSCAN model for clustering (the model includes PCA with 2 components)
dbscan = Clusterer(is_pca = True, clusterer = 'DBSCAN')
dbscan.fit(X, save = True)

# Score values for various eps values
dbscan.scores

# Preprocessed data (after scaling and PCA)
dbscan.X

# Prediction (basis on fit results)
preds = dbscan.predict(eps = 0.36, save = True)

import pandas as pd
from sklearn.datasets import load_iris
iris = pd.DataFrame(load_iris().data)

from plantstat import KNN

# Create and fit KNN with 5 neighbors
nn = KNN(5)
nn.fit(iris, save = True)

# Find neighbors for data subset
nn.find_neighbors(iris.iloc[:10, :], save = True)

# all kinds of data in the class
# nn.self_distances_
# nn.self_indices_
# nn.model_
# nn.all_data_

# nn.distances
# nn.indices

from plantstat.vision.stomata_vision import OpenStomataPredictor

predictor = OpenStomataPredictor('PATH', batch_size = 16)

predictor.predict(save = True)
predictor.visualize(save = True)

predictor.report_

predictor.test_img_paths_
predictor.test_preds_
predictor.test_classes_