def support_vector_regressor(X_train, X_test, y_train, y_test):
"""
"""
model = SVR()
model.fit(X_train, y_train)
y_pred = model.predict(X_test)
if y_test is not None:
return y_pred, \
rsquared_score(y_test, y_pred), \
rmse(y_test, y_pred), \
mean_absolute_percentage_error(y_test, y_pred), \
adj_rsquared_score(y_test, y_pred)
else:
return "There is no y_test to return the outputs", y_pred
def knn_regressor(X_train, X_test, y_train, y_test, n_neighbors=None):
"""
"""
model = KNeighborsRegressor(n_neighbors=n_neighbors)
model.fit(X_train, y_train)
y_pred = model.predict(X_test)
if y_test is not None:
return y_pred, \
rsquared_score(y_test, y_pred), \
rmse(y_test, y_pred), \
mean_absolute_percentage_error(y_test, y_pred), \
adj_rsquared_score(y_test, y_pred)
else:
return "There is no y_test to return the outputs", y_pred
def linear_regression(X_train=None, X_test=None, y_train=None, y_test=None,
samples_size=None, number_of_independent_columns=None):
"""
"""
lr = LinearRegression()
lr_model = lr.fit(X_train, y_train)
y_pred = lr_model.predict(X_test)
if y_test is not None:
return y_pred, \
rsquared_score(y_test, y_pred), \
rmse(y_test, y_pred), \
mean_absolute_percentage_error(y_test, y_pred), \
adj_rsquared_score(y_test, y_pred, n=samples_size,
p=number_of_independent_columns)
else:
return "There is no y_test to return the outputs", y_pred
def decision_tree_regressor(X_train=None, X_test=None, y_train=None, y_test=None,
max_depth=None, min_samples_leaf=None,
criterion=None, random_state=None):
"""
"""
dtr = DecisionTreeRegressor(max_depth=max_depth,
min_samples_leaf=min_samples_leaf, criterion=criterion,
random_state=random_state)
dtr_model = dtr.fit(X_train, y_train)
y_pred = dtr_model.predict(X_test)
if y_test is not None:
return y_pred, \
rsquared_score(y_test, y_pred), \
rmse(y_test, y_pred), \
mean_absolute_percentage_error(y_test, y_pred), \
adj_rsquared_score(y_test, y_pred)
else:
return "There is no y_test to return the outputs", y_pred
def random_forest_regression(X_train=None, X_test=None, y_train=None, y_test=None,
n_estimators=None, max_depth=None,
min_samples_leaf=None, criterion=None, random_state=None,
samples_size = None, number_of_independent_columns = None):
"""
"""
random_forest = RandomForestRegressor(n_estimators=n_estimators, max_depth=max_depth,
min_samples_leaf=min_samples_leaf, criterion=criterion,
random_state=random_state)
model = random_forest.fit(X_train, y_train)
y_pred = model.predict(X_test)
if y_test is not None:
return y_pred, \
rsquared_score(y_test, y_pred), \
rmse(y_test, y_pred), \
mean_absolute_percentage_error(y_test, y_pred), \
adj_rsquared_score(y_test, y_pred,
n=samples_size,
p=number_of_independent_columns)
else:
return "There is no y_test to return the outputs", y_pred
def evaluate(predicted,
actual,
method=None,
figsize=None,
ax=None,
plots=True,
reports=True,
samples_size=None,
number_of_independent_columns=None):
"""
:param actual: Your y_true
:param predicted:Your y_pred
:param figsize:Figure size
:param ax:axes
:param plots:True if you want all plots and is true by default
:param method: Classification or regression
:param reports: True by default and if you want reports you need to leave it as True
:param samples_size: Size of the sample and also to be used only for adj_r2_score
:param number_of_independent_columns: Number of columns which are independent and can only for adj_r2_score
:return: Returns all the classification metrics
"""
# Classification
if len(actual) != len(predicted):
raise exceptions.MetricValueError(
"The length of actual {0} and length of "
"predicted {1} are not the same".format(len(actual),
len(predicted)))
elif method == 'Classification':
print(
"\n---------------------\033[1mReports\033[0m-------------------------\n"
)
print("\033[1mAccuracy score:\033[0m {}".format(
accuracy(actual, predicted)))
print("\033[1mRecall score:\033[0m {}".format(recall(
actual, predicted)))
print("\033[1mPrecision score:\033[0m {}".format(
precision(actual, predicted)))
print("\033[1mF1 score:\033[0m {}".format(f1_score(actual, predicted)))
print(
"\n---------------------\033[1mPlots\033[0m-------------------------\n"
)
print("\033[1mClassification report:\033[0m: ")
classification_report_plot(actual, predicted, figsize=figsize, ax=ax)
print("\033[1mConfusion Matrix:\033[0m: ")
confusion_matrix_plot(actual, predicted, figsize, ax)
print("\033[1mROC curve:\033[0m ")
roc_auc_curve(actual, predicted, figsize, ax)
# Regression
if method == 'Regression':
actual = np.array(actual)
predicted = np.array(predicted)
print(
"\n-----------------------\033[1mReports\033[0m------------------------\n"
)
print("\033[1mRMSE:\033[0m {}".format(rmse(actual, predicted)))
print("\033[1mMAPE:\033[0m {}".format(
mean_absolute_percentage_error(actual, predicted)))
print("\033[1mR2 score:\033[0m {}".format(
rsquared_score(actual, predicted)))
print("\033[1mAdj R2 score:\033[0m {}".format(
adj_rsquared_score(actual, predicted, samples_size,
number_of_independent_columns)))
print(
"\n---------------------\033[1mPlots\033[0m-------------------------\n"
)
print("\033[1mRegression Plot 1:\033[0m ")
regression_line(actual, predicted)
print("\033[1mRegression Plot 2:\033[0m ")
sns_reg_plot(actual, predicted)