Add processing algorithm for balance data (SMOTETomek)

eis_qgis_plugin/eis_processing/algorithms/training_data_tools/balance_data.py

@@ -0,0 +1,87 @@
+from qgis.core import (
+    QgsProcessing,
+    QgsProcessingParameterEnum,
+    QgsProcessingParameterMapLayer,
+    QgsProcessingParameterMultipleLayers,
+    QgsProcessingParameterNumber,
+    QgsProcessingParameterRasterDestination,
+)
+
+from eis_qgis_plugin.eis_processing.eis_processing_algorithm import EISProcessingAlgorithm
+
+
+class EISBalanceData(EISProcessingAlgorithm):
+    def __init__(self) -> None:
+        super().__init__()
+
+        self._name = "balance_data"
+        self._display_name = "Balance data"
+        self._group = "Training data tools"
+        self._group_id = "training_data_tools"
+        self._short_help_string = """
+            Balances the classes of input dataset using SMOTETomek resampling method.
+
+            For more information about Imblearn SMOTETomek read the documentation here:
+            https://imbalanced-learn.org/stable/references/generated/imblearn.combine.SMOTETomek.html.
+        """
+
+    def initAlgorithm(self, config=None):
+        self.alg_parameters = [
+            "input_rasters",
+            "target_labels",
+            "sampling_strategy",
+            "random_state",
+            "output_raster",
+            "output_labels"
+        ]
+
+        input_raster_param = QgsProcessingParameterMultipleLayers(
+            name=self.alg_parameters[0],
+            description="Input data",
+            layerType=QgsProcessing.TypeRaster,
+        )
+        input_raster_param.setHelp("Input data to be resampled.")
+        self.addParameter(input_raster_param)
+
+        target_labels_param = QgsProcessingParameterMapLayer(
+            name=self.alg_parameters[1],
+            description="Target labels"
+        )
+        target_labels_param.setHelp("Labels corresponding to input data.")
+        self.addParameter(target_labels_param)
+
+        sampling_strategy_param = QgsProcessingParameterEnum(
+            name=self.alg_parameters[2],
+            description="Sampling strategy",
+            options=["minority", "not minority", "not majority", "all", "auto"],
+            defaultValue="auto"
+        )
+        sampling_strategy_param.setHelp(
+            "Sampling strategy to use. 'minority' resamples only the minority class, 'not minority' \
+            resamples all classes but the minority class, 'not majority' resamples all classes \
+            but the majority class, 'all' resamples all classes and 'auto' is equivalent to 'not majority'."
+        )
+        self.addParameter(sampling_strategy_param)
+
+        random_state_param = QgsProcessingParameterNumber(
+            name = self.alg_parameters[3],
+            description="Random state",
+            optional=True,
+            minValue=0
+        )
+        random_state_param.setHelp("Seed for random number generation.")
+        self.addParameter(random_state_param)
+
+        output_raster_param = QgsProcessingParameterRasterDestination(
+            name=self.alg_parameters[4],
+            description="Output raster"
+        )
+        output_raster_param.setHelp("Resampled feature data.")
+        self.addParameter(output_raster_param)
+
+        output_labels_param = QgsProcessingParameterRasterDestination(
+            name=self.alg_parameters[5],
+            description="Output labels"
+        )
+        output_labels_param.setHelp("Labels corresponding to resampled feature data.")
+        self.addParameter(output_labels_param)

eis_qgis_plugin/eis_processing/eis_provider.py

@@ -32,6 +32,7 @@ def loadAlgorithms(self) -> None:
         raster_processing = self.load_algorithms_from_directory("raster_processing")
         exploratory_analysis = self.load_algorithms_from_directory("exploratory_analysis")
         prediction = self.load_algorithms_from_directory("prediction")
+        training_data_tools = self.load_algorithms_from_directory("training_data_tools")
         transformations = self.load_algorithms_from_directory("transformations")
         transformations_coda = self.load_algorithms_from_directory("transformations_coda")
         utilities = self.load_algorithms_from_directory("utilities")
@@ -41,7 +42,7 @@ def loadAlgorithms(self) -> None:
         for algorithm in (
             evaluation + vector_processing + raster_processing +
             exploratory_analysis + prediction + transformations + utilities +
-            filtering + transformations_coda
+            filtering + transformations_coda + training_data_tools
         ):
             self.addAlgorithm(algorithm)
 

eis_qgis_plugin/eis_wizard/modeling/machine_learning/data_preparation.py

@@ -70,5 +70,5 @@ def __init__(self, parent, model_main) -> None:
         self.open_single_ilr_btn.clicked.connect(lambda _: processing.execAlgorithmDialog('eis:single_ilr_transform'))
         self.open_single_ilr_btn.setIcon(QIcon(QgsApplication.getThemeIcon("processingAlgorithm.svg")))
 
-        # self.open_balance_data_btn.clicked.connect(lambda _: processing.execAlgorithmDialog('eis:balance_data'))
+        self.open_balance_data_btn.clicked.connect(lambda _: processing.execAlgorithmDialog('eis:balance_data'))
         self.open_balance_data_btn.setIcon(QIcon(QgsApplication.getThemeIcon("processingAlgorithm.svg")))

eis_qgis_plugin/resources/ui/modeling/data_preparation.ui

@@ -334,7 +334,7 @@
             <item row="0" column="0">
              <widget class="QLabel" name="label_2">
               <property name="enabled">
-               <bool>false</bool>
+               <bool>true</bool>
               </property>
               <property name="text">
                <string>Balance data</string>
@@ -344,7 +344,7 @@
             <item row="0" column="1">
              <widget class="QPushButton" name="open_balance_data_btn">
               <property name="enabled">
-               <bool>false</bool>
+               <bool>true</bool>
               </property>
               <property name="text">
                <string>Open</string>