From 533334548eda829df9d57898945efec86dea9ebe Mon Sep 17 00:00:00 2001
From: Dreamlone <mik_sar@mail.ru>
Date: Fri, 3 Sep 2021 15:55:42 +0300
Subject: [PATCH 1/6] fitted values added
---
examples/time_series/ts_fitted_values.py | 38 ++++++++++++++++++++++++
fedot/core/pipelines/ts_wrappers.py | 36 +++++++++++++++++++++-
2 files changed, 73 insertions(+), 1 deletion(-)
create mode 100644 examples/time_series/ts_fitted_values.py
diff --git a/examples/time_series/ts_fitted_values.py b/examples/time_series/ts_fitted_values.py
new file mode 100644
index 000000000..8704fb053
--- /dev/null
+++ b/examples/time_series/ts_fitted_values.py
@@ -0,0 +1,38 @@
+import numpy as np
+import pandas as pd
+
+from matplotlib import pyplot as plt
+
+from fedot.core.data.data import InputData
+from fedot.core.pipelines.ts_wrappers import fitted_values
+from fedot.core.repository.tasks import Task, TaskTypesEnum, TsForecastingParams
+from fedot.core.pipelines.node import PrimaryNode, SecondaryNode
+from fedot.core.pipelines.pipeline import Pipeline
+from examples.time_series.ts_forecasting_tuning import get_complex_pipeline
+
+
+def show_fitted_time_series(len_forecast=24):
+ task = Task(TaskTypesEnum.ts_forecasting,
+ TsForecastingParams(forecast_length=len_forecast))
+
+ ts_input = InputData.from_csv_time_series(file_path='../../cases/data/time_series/metocean.csv',
+ task=task, target_column='value')
+
+ node_lagged = PrimaryNode('lagged')
+ node_lagged.custom_params = {'window_size': 4}
+ node_final = SecondaryNode('ridge', nodes_from=[node_lagged])
+ pipeline = Pipeline(node_final)
+ train_predicted = pipeline.fit(ts_input)
+
+ fitted_ts_10 = fitted_values(train_predicted, 10)
+ fitted_ts_act = fitted_values(train_predicted)
+ plt.plot(ts_input.idx, ts_input.target, label='Actual time series')
+ plt.plot(fitted_ts_10.idx, fitted_ts_10.predict, label='Fitted values horizon 10')
+ plt.plot(fitted_ts_act.idx, fitted_ts_act.predict, label='Fitted values all')
+ plt.legend()
+ plt.grid()
+ plt.show()
+
+
+if __name__ == '__main__':
+ show_fitted_time_series()
diff --git a/fedot/core/pipelines/ts_wrappers.py b/fedot/core/pipelines/ts_wrappers.py
index 9ab1f6580..cc0fe2eb0 100644
--- a/fedot/core/pipelines/ts_wrappers.py
+++ b/fedot/core/pipelines/ts_wrappers.py
@@ -1,9 +1,11 @@
+from copy import copy
from typing import Union
import numpy as np
-from fedot.core.data.data import InputData
+from fedot.core.data.data import InputData, OutputData
from fedot.core.data.multi_modal import MultiModalData
+from fedot.core.operations.evaluation.operation_implementations.data_operations.ts_transformations import _ts_to_table
from fedot.core.repository.dataset_types import DataTypesEnum
from fedot.core.repository.tasks import TaskTypesEnum
@@ -162,6 +164,38 @@ def in_sample_ts_forecast(pipeline, input_data: Union[InputData, MultiModalData]
return final_forecast
+def fitted_values(train_predicted: OutputData, horizon_step: int = None):
+ """ The method converts a multidimensional lagged array into an
+ one-dimensional array - time series
+
+ :param train_predicted: OutputData
+ :param horizon_step: index of elements for forecast
+ """
+ copied_data = copy(train_predicted)
+ if horizon_step is not None:
+ # Take particular forecast step
+ copied_data.predict = copied_data.predict[:, horizon_step]
+ copied_data.idx = copied_data.idx + horizon_step
+ return copied_data
+ else:
+ # Perform collapse with averaging
+ forecast_length = copied_data.task.task_params.forecast_length
+ # LMatrix with indices in cells
+ indices_range = np.arange(0, len(copied_data.predict))
+ _, idx_matrix = _ts_to_table(idx=indices_range,
+ time_series=indices_range,
+ window_size=forecast_length)
+ predicted_matrix = copied_data.predict
+ predicted_matrix = predicted_matrix[-len(idx_matrix):, :]
+ final_predictions = []
+ for index in indices_range:
+ vals = predicted_matrix[idx_matrix == index]
+ mean_value = np.mean(vals)
+ final_predictions.append(mean_value)
+ copied_data.predict = np.array(final_predictions)
+ return copied_data
+
+
def _calculate_number_of_steps(scope_len, horizon):
""" Method return amount of iterations which must be done for multistep
time series forecasting
--
GitLab
From 1ea6472dd2d4b28b64938c43b61ff288d20e95ad Mon Sep 17 00:00:00 2001
From: Dreamlone <mik_sar@mail.ru>
Date: Fri, 3 Sep 2021 18:04:45 +0300
Subject: [PATCH 2/6] improve cutting
---
...ts_fitted_values.py => ts_fitted_target.py} | 18 +++++-------------
.../data_operations/ts_transformations.py | 3 ++-
fedot/core/pipelines/ts_wrappers.py | 12 ++++++++----
.../pipelines/test_pipeline_ts_wrappers.py | 14 ++++++++++++++
4 files changed, 29 insertions(+), 18 deletions(-)
rename examples/time_series/{ts_fitted_values.py => ts_fitted_target.py} (60%)
diff --git a/examples/time_series/ts_fitted_values.py b/examples/time_series/ts_fitted_target.py
similarity index 60%
rename from examples/time_series/ts_fitted_values.py
rename to examples/time_series/ts_fitted_target.py
index 8704fb053..936fd1a00 100644
--- a/examples/time_series/ts_fitted_values.py
+++ b/examples/time_series/ts_fitted_target.py
@@ -1,14 +1,9 @@
-import numpy as np
-import pandas as pd
-
from matplotlib import pyplot as plt
from fedot.core.data.data import InputData
-from fedot.core.pipelines.ts_wrappers import fitted_values
+from fedot.core.pipelines.ts_wrappers import fitted_target
from fedot.core.repository.tasks import Task, TaskTypesEnum, TsForecastingParams
-from fedot.core.pipelines.node import PrimaryNode, SecondaryNode
-from fedot.core.pipelines.pipeline import Pipeline
-from examples.time_series.ts_forecasting_tuning import get_complex_pipeline
+from test.unit.pipelines.test_pipeline_ts_wrappers import get_simple_short_lagged_pipeline
def show_fitted_time_series(len_forecast=24):
@@ -18,14 +13,11 @@ def show_fitted_time_series(len_forecast=24):
ts_input = InputData.from_csv_time_series(file_path='../../cases/data/time_series/metocean.csv',
task=task, target_column='value')
- node_lagged = PrimaryNode('lagged')
- node_lagged.custom_params = {'window_size': 4}
- node_final = SecondaryNode('ridge', nodes_from=[node_lagged])
- pipeline = Pipeline(node_final)
+ pipeline = get_simple_short_lagged_pipeline()
train_predicted = pipeline.fit(ts_input)
- fitted_ts_10 = fitted_values(train_predicted, 10)
- fitted_ts_act = fitted_values(train_predicted)
+ fitted_ts_10 = fitted_target(train_predicted, 10)
+ fitted_ts_act = fitted_target(train_predicted)
plt.plot(ts_input.idx, ts_input.target, label='Actual time series')
plt.plot(fitted_ts_10.idx, fitted_ts_10.predict, label='Fitted values horizon 10')
plt.plot(fitted_ts_act.idx, fitted_ts_act.predict, label='Fitted values all')
diff --git a/fedot/core/operations/evaluation/operation_implementations/data_operations/ts_transformations.py b/fedot/core/operations/evaluation/operation_implementations/data_operations/ts_transformations.py
index fbb624593..c3ed2dfa0 100644
--- a/fedot/core/operations/evaluation/operation_implementations/data_operations/ts_transformations.py
+++ b/fedot/core/operations/evaluation/operation_implementations/data_operations/ts_transformations.py
@@ -317,8 +317,9 @@ def _ts_to_table(idx, time_series: np.array, window_size: int):
transformed = np.array(lagged_dataframe)
- # Generate dataset with features
+ # Remove extra column (it will go to target)
features_columns = transformed[:, 1:]
+ # Generate dataset with features
features_columns = np.fliplr(features_columns)
# First n elements in time series are removed
diff --git a/fedot/core/pipelines/ts_wrappers.py b/fedot/core/pipelines/ts_wrappers.py
index cc0fe2eb0..72766de42 100644
--- a/fedot/core/pipelines/ts_wrappers.py
+++ b/fedot/core/pipelines/ts_wrappers.py
@@ -164,7 +164,7 @@ def in_sample_ts_forecast(pipeline, input_data: Union[InputData, MultiModalData]
return final_forecast
-def fitted_values(train_predicted: OutputData, horizon_step: int = None):
+def fitted_target(train_predicted: OutputData, horizon_step: int = None):
""" The method converts a multidimensional lagged array into an
one-dimensional array - time series
@@ -180,19 +180,23 @@ def fitted_values(train_predicted: OutputData, horizon_step: int = None):
else:
# Perform collapse with averaging
forecast_length = copied_data.task.task_params.forecast_length
- # LMatrix with indices in cells
- indices_range = np.arange(0, len(copied_data.predict))
+ # Extend source index range
+ indices_range = np.arange(copied_data.idx[0], copied_data.idx[-1] + forecast_length + 1)
+ # Lagged matrix with indices in cells
_, idx_matrix = _ts_to_table(idx=indices_range,
time_series=indices_range,
window_size=forecast_length)
predicted_matrix = copied_data.predict
- predicted_matrix = predicted_matrix[-len(idx_matrix):, :]
+
+ # For every index calculate mean predictions (by all forecast steps)
final_predictions = []
+ indices_range = indices_range[:-1]
for index in indices_range:
vals = predicted_matrix[idx_matrix == index]
mean_value = np.mean(vals)
final_predictions.append(mean_value)
copied_data.predict = np.array(final_predictions)
+ copied_data.idx = indices_range
return copied_data
diff --git a/test/unit/pipelines/test_pipeline_ts_wrappers.py b/test/unit/pipelines/test_pipeline_ts_wrappers.py
index 4362a8449..083c0735b 100644
--- a/test/unit/pipelines/test_pipeline_ts_wrappers.py
+++ b/test/unit/pipelines/test_pipeline_ts_wrappers.py
@@ -7,6 +7,7 @@ from fedot.core.pipelines.pipeline import Pipeline
from fedot.core.pipelines.ts_wrappers import in_sample_ts_forecast, out_of_sample_ts_forecast
from fedot.core.repository.dataset_types import DataTypesEnum
from fedot.core.repository.tasks import Task, TaskTypesEnum, TsForecastingParams
+from fedot.core.pipelines.ts_wrappers import fitted_target
def prepare_input_data(forecast_length, horizon):
@@ -88,3 +89,16 @@ def test_in_sample_ts_forecast_correct():
is_forecast_correct = True
assert is_forecast_correct
+
+
+def test_fitted_target_correct():
+ from matplotlib import pyplot as plt
+ simple_length, multi_length = 5, 5
+ ts_input, _ = prepare_input_data(simple_length, multi_length)
+
+ pipeline = get_simple_short_lagged_pipeline()
+ train_predicted = pipeline.fit(ts_input)
+
+ fitted_ts_values = fitted_target(train_predicted)
+
+ assert len(fitted_ts_values.predict) == len(ts_input.target) - 4
--
GitLab
From eb23300cdf983bd0f5cca8b40bf9dd129976c6e8 Mon Sep 17 00:00:00 2001
From: Dreamlone <mik_sar@mail.ru>
Date: Fri, 3 Sep 2021 19:14:17 +0300
Subject: [PATCH 3/6] in-sample fitted
---
examples/time_series/ts_fitted_target.py | 5 ++++-
fedot/core/pipelines/ts_wrappers.py | 18 +++++++++++++++---
2 files changed, 19 insertions(+), 4 deletions(-)
diff --git a/examples/time_series/ts_fitted_target.py b/examples/time_series/ts_fitted_target.py
index 936fd1a00..312343c0a 100644
--- a/examples/time_series/ts_fitted_target.py
+++ b/examples/time_series/ts_fitted_target.py
@@ -1,7 +1,7 @@
from matplotlib import pyplot as plt
from fedot.core.data.data import InputData
-from fedot.core.pipelines.ts_wrappers import fitted_target
+from fedot.core.pipelines.ts_wrappers import fitted_target, in_sample_fitted_target
from fedot.core.repository.tasks import Task, TaskTypesEnum, TsForecastingParams
from test.unit.pipelines.test_pipeline_ts_wrappers import get_simple_short_lagged_pipeline
@@ -18,6 +18,9 @@ def show_fitted_time_series(len_forecast=24):
fitted_ts_10 = fitted_target(train_predicted, 10)
fitted_ts_act = fitted_target(train_predicted)
+
+ in_sample_fitted_target(train_predicted)
+
plt.plot(ts_input.idx, ts_input.target, label='Actual time series')
plt.plot(fitted_ts_10.idx, fitted_ts_10.predict, label='Fitted values horizon 10')
plt.plot(fitted_ts_act.idx, fitted_ts_act.predict, label='Fitted values all')
diff --git a/fedot/core/pipelines/ts_wrappers.py b/fedot/core/pipelines/ts_wrappers.py
index 72766de42..4cf24716d 100644
--- a/fedot/core/pipelines/ts_wrappers.py
+++ b/fedot/core/pipelines/ts_wrappers.py
@@ -164,12 +164,13 @@ def in_sample_ts_forecast(pipeline, input_data: Union[InputData, MultiModalData]
return final_forecast
-def fitted_target(train_predicted: OutputData, horizon_step: int = None):
+def fitted_target(train_predicted: OutputData, horizon_step: int = None) -> OutputData:
""" The method converts a multidimensional lagged array into an
- one-dimensional array - time series
+ one-dimensional array - time series based on predicted values for training sample
:param train_predicted: OutputData
- :param horizon_step: index of elements for forecast
+ :param horizon_step: index of elements for forecast. If None - perform
+ averaging for all forecasting steps
"""
copied_data = copy(train_predicted)
if horizon_step is not None:
@@ -200,6 +201,17 @@ def fitted_target(train_predicted: OutputData, horizon_step: int = None):
return copied_data
+def in_sample_fitted_target(train_predicted: OutputData) -> OutputData:
+ """ Perform in sample validation based on training sample """
+ forecast_length = train_predicted.task.task_params.forecast_length
+ all_values = []
+ step = 0
+ while step < len(train_predicted.idx):
+ all_values.extend(train_predicted.predict[step, :])
+ step += forecast_length
+ # TODO finish implementation
+
+
def _calculate_number_of_steps(scope_len, horizon):
""" Method return amount of iterations which must be done for multistep
time series forecasting
--
GitLab
From 5dee71f4e0391f69ba8960852e000e23e1747163 Mon Sep 17 00:00:00 2001
From: Dreamlone <mik_sar@mail.ru>
Date: Tue, 7 Sep 2021 16:04:02 +0300
Subject: [PATCH 4/6] add in-sample fitted to test & examples
---
examples/time_series/ts_fitted_target.py | 19 +++++++++++-------
fedot/core/pipelines/ts_wrappers.py | 20 +++++++++++++++++--
.../pipelines/test_pipeline_ts_wrappers.py | 12 +++++------
3 files changed, 36 insertions(+), 15 deletions(-)
diff --git a/examples/time_series/ts_fitted_target.py b/examples/time_series/ts_fitted_target.py
index 312343c0a..f0889a74b 100644
--- a/examples/time_series/ts_fitted_target.py
+++ b/examples/time_series/ts_fitted_target.py
@@ -7,6 +7,10 @@ from test.unit.pipelines.test_pipeline_ts_wrappers import get_simple_short_lagge
def show_fitted_time_series(len_forecast=24):
+ """
+ Shows an example of how to get fitted values of a time series by any
+ pipeline created by FEDOT
+ """
task = Task(TaskTypesEnum.ts_forecasting,
TsForecastingParams(forecast_length=len_forecast))
@@ -16,18 +20,19 @@ def show_fitted_time_series(len_forecast=24):
pipeline = get_simple_short_lagged_pipeline()
train_predicted = pipeline.fit(ts_input)
- fitted_ts_10 = fitted_target(train_predicted, 10)
+ # Get fitted values
+ fitted_ts_10 = fitted_target(train_predicted, 1)
fitted_ts_act = fitted_target(train_predicted)
+ in_sample_validated = in_sample_fitted_target(train_predicted)
- in_sample_fitted_target(train_predicted)
-
- plt.plot(ts_input.idx, ts_input.target, label='Actual time series')
- plt.plot(fitted_ts_10.idx, fitted_ts_10.predict, label='Fitted values horizon 10')
- plt.plot(fitted_ts_act.idx, fitted_ts_act.predict, label='Fitted values all')
+ plt.plot(ts_input.idx, ts_input.target, label='Actual time series', alpha=0.8)
+ plt.plot(fitted_ts_10.idx, fitted_ts_10.predict, label='Fitted values horizon 10', alpha=0.2)
+ plt.plot(fitted_ts_act.idx, fitted_ts_act.predict, label='Fitted values all', alpha=0.2)
+ plt.plot(in_sample_validated.idx, in_sample_validated.predict, label='In-sample fitted values')
plt.legend()
plt.grid()
plt.show()
if __name__ == '__main__':
- show_fitted_time_series()
+ show_fitted_time_series(55)
diff --git a/fedot/core/pipelines/ts_wrappers.py b/fedot/core/pipelines/ts_wrappers.py
index 4cf24716d..e0b9874ae 100644
--- a/fedot/core/pipelines/ts_wrappers.py
+++ b/fedot/core/pipelines/ts_wrappers.py
@@ -181,8 +181,10 @@ def fitted_target(train_predicted: OutputData, horizon_step: int = None) -> Outp
else:
# Perform collapse with averaging
forecast_length = copied_data.task.task_params.forecast_length
+
# Extend source index range
indices_range = np.arange(copied_data.idx[0], copied_data.idx[-1] + forecast_length + 1)
+
# Lagged matrix with indices in cells
_, idx_matrix = _ts_to_table(idx=indices_range,
time_series=indices_range,
@@ -206,10 +208,24 @@ def in_sample_fitted_target(train_predicted: OutputData) -> OutputData:
forecast_length = train_predicted.task.task_params.forecast_length
all_values = []
step = 0
- while step < len(train_predicted.idx):
+ # Glues together parts of predictions using "in-sample" way
+ while step < len(train_predicted.predict):
all_values.extend(train_predicted.predict[step, :])
step += forecast_length
- # TODO finish implementation
+
+ # In some cases it doesn't reach the end
+ if not np.isclose(all_values[-1], train_predicted.predict[-1, -1]):
+ missing_part_index = step - len(train_predicted.predict) + 1
+ # Store missing predicted values
+ all_values.extend(train_predicted.predict[-1, missing_part_index:])
+
+ copied_data = copy(train_predicted)
+ copied_data.predict = np.array(all_values)
+ # Update indices
+ first_id = copied_data.idx[0]
+ copied_data.idx = np.arange(first_id, first_id + len(all_values))
+
+ return copied_data
def _calculate_number_of_steps(scope_len, horizon):
diff --git a/test/unit/pipelines/test_pipeline_ts_wrappers.py b/test/unit/pipelines/test_pipeline_ts_wrappers.py
index 083c0735b..56a5ef2e0 100644
--- a/test/unit/pipelines/test_pipeline_ts_wrappers.py
+++ b/test/unit/pipelines/test_pipeline_ts_wrappers.py
@@ -4,7 +4,7 @@ from sklearn.metrics import mean_absolute_error
from fedot.core.data.data import InputData
from fedot.core.pipelines.node import PrimaryNode, SecondaryNode
from fedot.core.pipelines.pipeline import Pipeline
-from fedot.core.pipelines.ts_wrappers import in_sample_ts_forecast, out_of_sample_ts_forecast
+from fedot.core.pipelines.ts_wrappers import in_sample_ts_forecast, out_of_sample_ts_forecast, in_sample_fitted_target
from fedot.core.repository.dataset_types import DataTypesEnum
from fedot.core.repository.tasks import Task, TaskTypesEnum, TsForecastingParams
from fedot.core.pipelines.ts_wrappers import fitted_target
@@ -92,13 +92,13 @@ def test_in_sample_ts_forecast_correct():
def test_fitted_target_correct():
- from matplotlib import pyplot as plt
simple_length, multi_length = 5, 5
ts_input, _ = prepare_input_data(simple_length, multi_length)
- pipeline = get_simple_short_lagged_pipeline()
- train_predicted = pipeline.fit(ts_input)
+ for fitted_func in [fitted_target, in_sample_fitted_target]:
+ pipeline = get_simple_short_lagged_pipeline()
+ train_predicted = pipeline.fit(ts_input)
- fitted_ts_values = fitted_target(train_predicted)
+ fitted_ts_values = fitted_func(train_predicted)
- assert len(fitted_ts_values.predict) == len(ts_input.target) - 4
+ assert len(fitted_ts_values.predict) == len(ts_input.target) - 4
--
GitLab
From b5f34acf3b8092ec3a352ee4acadc584e95030ec Mon Sep 17 00:00:00 2001
From: Dreamlone <mik_sar@mail.ru>
Date: Wed, 8 Sep 2021 11:39:27 +0300
Subject: [PATCH 5/6] add description
---
.../{ts_fitted_target.py => ts_fitted_values.py} | 12 +++++++++---
1 file changed, 9 insertions(+), 3 deletions(-)
rename examples/time_series/{ts_fitted_target.py => ts_fitted_values.py} (78%)
diff --git a/examples/time_series/ts_fitted_target.py b/examples/time_series/ts_fitted_values.py
similarity index 78%
rename from examples/time_series/ts_fitted_target.py
rename to examples/time_series/ts_fitted_values.py
index f0889a74b..3e4fd94b6 100644
--- a/examples/time_series/ts_fitted_target.py
+++ b/examples/time_series/ts_fitted_values.py
@@ -10,6 +10,10 @@ def show_fitted_time_series(len_forecast=24):
"""
Shows an example of how to get fitted values of a time series by any
pipeline created by FEDOT
+
+ fitted values - are the predictions of the pipelines on the training sample.
+ For time series, these values show how well the model reproduces the time
+ series structure
"""
task = Task(TaskTypesEnum.ts_forecasting,
TsForecastingParams(forecast_length=len_forecast))
@@ -20,9 +24,11 @@ def show_fitted_time_series(len_forecast=24):
pipeline = get_simple_short_lagged_pipeline()
train_predicted = pipeline.fit(ts_input)
- # Get fitted values
- fitted_ts_10 = fitted_target(train_predicted, 1)
+ # Get fitted values for every 10th forecast
+ fitted_ts_10 = fitted_target(train_predicted, 10)
+ # Average for all forecasting horizons
fitted_ts_act = fitted_target(train_predicted)
+ # In-sample forecasting fitted values
in_sample_validated = in_sample_fitted_target(train_predicted)
plt.plot(ts_input.idx, ts_input.target, label='Actual time series', alpha=0.8)
@@ -35,4 +41,4 @@ def show_fitted_time_series(len_forecast=24):
if __name__ == '__main__':
- show_fitted_time_series(55)
+ show_fitted_time_series()
--
GitLab
From 0545eb4f4e1e6ba1560a449429bfccd747205139 Mon Sep 17 00:00:00 2001
From: Dreamlone <mik_sar@mail.ru>
Date: Wed, 8 Sep 2021 12:32:29 +0300
Subject: [PATCH 6/6] change names
---
examples/time_series/ts_fitted_values.py | 8 ++++----
fedot/core/pipelines/ts_wrappers.py | 4 ++--
test/unit/pipelines/test_pipeline_ts_wrappers.py | 8 ++++----
3 files changed, 10 insertions(+), 10 deletions(-)
diff --git a/examples/time_series/ts_fitted_values.py b/examples/time_series/ts_fitted_values.py
index 3e4fd94b6..01c4e0db0 100644
--- a/examples/time_series/ts_fitted_values.py
+++ b/examples/time_series/ts_fitted_values.py
@@ -1,7 +1,7 @@
from matplotlib import pyplot as plt
from fedot.core.data.data import InputData
-from fedot.core.pipelines.ts_wrappers import fitted_target, in_sample_fitted_target
+from fedot.core.pipelines.ts_wrappers import fitted_values, in_sample_fitted_values
from fedot.core.repository.tasks import Task, TaskTypesEnum, TsForecastingParams
from test.unit.pipelines.test_pipeline_ts_wrappers import get_simple_short_lagged_pipeline
@@ -25,11 +25,11 @@ def show_fitted_time_series(len_forecast=24):
train_predicted = pipeline.fit(ts_input)
# Get fitted values for every 10th forecast
- fitted_ts_10 = fitted_target(train_predicted, 10)
+ fitted_ts_10 = fitted_values(train_predicted, 10)
# Average for all forecasting horizons
- fitted_ts_act = fitted_target(train_predicted)
+ fitted_ts_act = fitted_values(train_predicted)
# In-sample forecasting fitted values
- in_sample_validated = in_sample_fitted_target(train_predicted)
+ in_sample_validated = in_sample_fitted_values(train_predicted)
plt.plot(ts_input.idx, ts_input.target, label='Actual time series', alpha=0.8)
plt.plot(fitted_ts_10.idx, fitted_ts_10.predict, label='Fitted values horizon 10', alpha=0.2)
diff --git a/fedot/core/pipelines/ts_wrappers.py b/fedot/core/pipelines/ts_wrappers.py
index e0b9874ae..dea6d853e 100644
--- a/fedot/core/pipelines/ts_wrappers.py
+++ b/fedot/core/pipelines/ts_wrappers.py
@@ -164,7 +164,7 @@ def in_sample_ts_forecast(pipeline, input_data: Union[InputData, MultiModalData]
return final_forecast
-def fitted_target(train_predicted: OutputData, horizon_step: int = None) -> OutputData:
+def fitted_values(train_predicted: OutputData, horizon_step: int = None) -> OutputData:
""" The method converts a multidimensional lagged array into an
one-dimensional array - time series based on predicted values for training sample
@@ -203,7 +203,7 @@ def fitted_target(train_predicted: OutputData, horizon_step: int = None) -> Outp
return copied_data
-def in_sample_fitted_target(train_predicted: OutputData) -> OutputData:
+def in_sample_fitted_values(train_predicted: OutputData) -> OutputData:
""" Perform in sample validation based on training sample """
forecast_length = train_predicted.task.task_params.forecast_length
all_values = []
diff --git a/test/unit/pipelines/test_pipeline_ts_wrappers.py b/test/unit/pipelines/test_pipeline_ts_wrappers.py
index 56a5ef2e0..23e9713fb 100644
--- a/test/unit/pipelines/test_pipeline_ts_wrappers.py
+++ b/test/unit/pipelines/test_pipeline_ts_wrappers.py
@@ -4,10 +4,10 @@ from sklearn.metrics import mean_absolute_error
from fedot.core.data.data import InputData
from fedot.core.pipelines.node import PrimaryNode, SecondaryNode
from fedot.core.pipelines.pipeline import Pipeline
-from fedot.core.pipelines.ts_wrappers import in_sample_ts_forecast, out_of_sample_ts_forecast, in_sample_fitted_target
+from fedot.core.pipelines.ts_wrappers import in_sample_ts_forecast, out_of_sample_ts_forecast, in_sample_fitted_values
from fedot.core.repository.dataset_types import DataTypesEnum
from fedot.core.repository.tasks import Task, TaskTypesEnum, TsForecastingParams
-from fedot.core.pipelines.ts_wrappers import fitted_target
+from fedot.core.pipelines.ts_wrappers import fitted_values
def prepare_input_data(forecast_length, horizon):
@@ -91,11 +91,11 @@ def test_in_sample_ts_forecast_correct():
assert is_forecast_correct
-def test_fitted_target_correct():
+def test_fitted_values_correct():
simple_length, multi_length = 5, 5
ts_input, _ = prepare_input_data(simple_length, multi_length)
- for fitted_func in [fitted_target, in_sample_fitted_target]:
+ for fitted_func in [fitted_values, in_sample_fitted_values]:
pipeline = get_simple_short_lagged_pipeline()
train_predicted = pipeline.fit(ts_input)
--
GitLab