v1.8.1

go_server/src/utils.go

@@ -239,7 +239,8 @@ func copyOutput(r io.Reader, response *string) {
 
 // ReadFile reads a file and returns its content as a string
 func ReadFile(filename string) string {
-	absPath, _ := filepath.Abs(filename)
+	cleanPath := strings.Trim(filename, "\" \t\n\r")
+	absPath, _ := filepath.Abs(cleanPath)
 	log.Println("Reading file: " + absPath)
 	data, err := os.ReadFile(absPath)
 	if err != nil {

pythonCode/med_libs/MEDml/nodes/ModelHandler.py

@@ -1,3 +1,4 @@
+import ast
 import copy
 import json
 from typing import Union
@@ -131,6 +132,7 @@ def __calculate_all_metrics(self, y_true, y_pred, y_pred_proba=None):
             metrics['MCC'] = round(matthews_corrcoef(y_true, y_pred), 3)
 
         except Exception as e:
+            raise ValueError(f"Error calculating metrics: {e}")
             print(f"Error calculating metrics: {e}")
             # Set default values for all metrics
             default_metrics = ["AUC", "Sensitivity", "Specificity", "PPV", "NPV", "Accuracy", "F1", "MCC"]
@@ -145,7 +147,7 @@ def __calculate_overall_metrics(self, fold_metrics):
         log_metrics = {}
 
         if not fold_metrics:
-            return overall_metrics
+            return overall_metrics, log_metrics
 
         # Get all metric names from first fold
         first_fold_metrics = list(fold_metrics.values())[0]
@@ -401,7 +403,7 @@ def __custom_train_and_evaluate(
                 if self.optimize_threshold:
                     if len(pycaret_exp.get_config('y').unique()) == 2 and not self.ensembleEnabled:
                         self.CodeHandler.add_line("code", f"# Optimizing model threshold based on {self.threshold_optimization_metric}", indent=0)
-                        self.CodeHandler.add_line("code", f"best_model = pycaret_exp.optimize_threshold(best_model, metric='{self.threshold_optimization_metric}')", indent=0)
+                        self.CodeHandler.add_line("code", f"best_model = pycaret_exp.optimize_threshold(best_model, optimize='{self.threshold_optimization_metric}')", indent=0)
 
                 # Finalize the model
                 if finalize:
@@ -552,7 +554,7 @@ def __handle_splitted_data(self, experiment: dict, settings: dict, **kwargs) ->
                     )
                     self.CodeHandler.add_line(
                         "code",
-                        f"trained_models = [pycaret_exp.optimize_threshold(trained_models[0], metric='{self.threshold_optimization_metric}')]"
+                        f"trained_models = [pycaret_exp.optimize_threshold(trained_models[0], optimize='{self.threshold_optimization_metric}')]"
                     )
 
             if finalize:
@@ -756,7 +758,7 @@ def _execute(self, experiment: dict = None, **kwargs) -> json:
 
                 else:
                     trained_models = [experiment['pycaret_exp'].optimize_threshold(trained_models[0], optimize=self.threshold_optimization_metric)]
-                    self.CodeHandler.add_line("code", f"trained_models = [pycaret_exp.optimize_threshold(trained_models[0], metric='{self.threshold_optimization_metric}')]")
+                    self.CodeHandler.add_line("code", f"trained_models = [pycaret_exp.optimize_threshold(trained_models[0], optimize='{self.threshold_optimization_metric}')]")
 
             if finalize:
                 trained_models = [experiment['pycaret_exp'].finalize_model(model) for model in trained_models]

pythonCode/med_libs/MEDml/nodes/ModelIO.py

@@ -58,13 +58,14 @@ def _execute(self, experiment: dict = None, **kwargs) -> json:
                 if dir(fitted_model).__contains__('feature_names_in_'):
                     model_features = fitted_model.__getattribute__('feature_names_in_')
                 elif dir(fitted_model).__contains__('feature_name_') and model_features is None:
-                    model_features = fitted_model.__getattribute__('feature_name_')
+                    model_features = fitted_model.__getattribute__('feature_names_in_')
+                elif dir(fitted_model).__contains__('classifier_') and dir(fitted_model.classifier_).__contains__('feature_names_in_'):
+                    model_features = fitted_model.classifier_.feature_names_in_
                 else:
-                    model_features=  fitted_model.__getattr__('feature_names_in_')
-
+                    raise ValueError(f"Could not find model features. Model attributes : {dir(fitted_model)}, model type: {type(fitted_model)}")
                 if model_features is None:
-                    raise ValueError(f"Could not find model features. Model attributes : {dir(fitted_model)}, model type: {type(fitted_model)}, model features: {model_features}")
-                
+                    raise ValueError(f"Could not find model features. Model attributes : {dir(fitted_model)}, model type: {type(fitted_model)}")
+
                 model_features = list(model_features)
 
                 # Model's name

pythonCode/med_libs/MEDml/nodes/Split.py

@@ -269,7 +269,7 @@ def _execute(self, experiment: dict = None, **kwargs) -> json:
                 splitter = StratifiedKFold(n_splits=cv_folds, shuffle=True, random_state=random_state)
                 self.CodeHandler.add_line("code", f"splitter = StratifiedKFold(n_splits={cv_folds}, shuffle=True, random_state={random_state})")
                 fold_iter = splitter.split(np.zeros(n_samples), y)
-                self.CodeHandler.add_line("code", f"fold_iter = splitter.split(np.zeros({n_samples}), y)")
+                self.CodeHandler.add_line("code", f"fold_iter = splitter.split(np.zeros(len(dataset)), y)")
             else:
                 splitter = KFold(n_splits=cv_folds, shuffle=True, random_state=random_state)
                 self.CodeHandler.add_line("code", f"splitter = KFold(n_splits={cv_folds}, shuffle=True, random_state={random_state})")

pythonCode/modules/superset/SupersetEnvManager.py

@@ -41,7 +41,21 @@ def _get_build_env(self):
         """
 
         env = os.environ.copy()
-        if sys.platform != "win32":
+        if sys.platform == "darwin":
+            # Force the compiler to use the system SDK and headers
+            cc = shutil.which("clang") or shutil.which("gcc")
+            cxx = shutil.which("clang++") or shutil.which("g++")
+            if cc:
+                env["CC"] = cc
+            if cxx:
+                env["CXX"] = cxx
+
+            # Critical for 'cryptography' and 'python-geohash' compilation
+            env["LDFLAGS"] = "-L/usr/local/opt/openssl/lib -L/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/lib"
+            env["CPPFLAGS"] = "-I/usr/local/opt/openssl/include -I/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include"
+            # Prevents error: 'implicit declaration of function' during geohash build
+            env["CFLAGS"] = "-Wno-error=implicit-function-declaration"
+        elif sys.platform != "win32":
             cc = shutil.which("gcc") or shutil.which("cc")
             cxx = shutil.which("g++") or shutil.which("c++")
             if cc:

pythonEnv/requirements.txt

@@ -243,7 +243,6 @@ safetensors==0.4.0
 schemdraw==0.15
 scikit-base==0.5.2
 scikit-image==0.19.3
-scikit-learn==1.2.0
 scikit-plot==0.3.7
 scipy==1.10.1
 seaborn==0.12.2

renderer/components/learning/nodesTypes/splitNode.jsx

@@ -254,7 +254,7 @@ const SplitNode = ({ id, data }) => {
       key={id}
       id={id}
       data={data}
-      color="#EAD196"
+      color="#cea037"
       setupParam={data.setupParam}
       nodeLink="/documentation/split"
       defaultSettings={

renderer/components/learning/nodesTypes/trainModelNode.jsx

@@ -30,11 +30,11 @@ const TrainModelNode = ({ id, data }) => {
   const [modalShowTuning, setModalShowTuning] = useState(false)
   const { updateNode } = useContext(FlowFunctionsContext)
   const [IntegrateTuning, setIntegrateTuning] = useState(data.internal.isTuningEnabled ?? false)
-  const [optimizeThresh, setOptimizeThresh] = useState(data.internal.isOptimizeThreshold ?? false)
-  const [ensembleEnabled, setEnsembleEnabled] = useState(data.internal.settings.isEnsembleEnabled ?? false)
-  const [calibrateEnabled, setCalibrateEnabled] = useState(data.internal.settings.isCalibrateEnabled ?? false)
+  const [optimizeThresh, setOptimizeThresh] = useState(data.internal.optimizeThreshold ?? false)
+  const [ensembleEnabled, setEnsembleEnabled] = useState(data.internal.ensembleEnabled ?? false)
+  const [calibrateEnabled, setCalibrateEnabled] = useState(data.internal.calibrateEnabled ?? false)
 
-  // Check if isTuningEnabled exists in data.internal, if not initialize it
+  // Check if default settings exists in data.internal, if not initialize it
   useEffect(() => {
     let hasUpdates = false
 
@@ -58,8 +58,9 @@ const TrainModelNode = ({ id, data }) => {
     const defaults = {
       isTuningEnabled: false,
       useTuningGrid: false,
-      isEnsembleEnabled: false,
-      isCalibrateEnabled: false,
+      ensembleEnabled: false,
+      calibrateEnabled: false,
+      optimizeThreshold: false,
       settingsEnsembling: {},
       settingsCalibration: {}
     }
@@ -185,21 +186,6 @@ const TrainModelNode = ({ id, data }) => {
     })
   }
 
-  /**
-   *
-   * @param {Object} e the event of the checkbox
-   * @description
-   * This function is used to handle the checkbox for enabling the tuning
-   */
-  const handleIntegration = (e) => {
-    setIntegrateTuning(e.value)
-    data.internal.isTuningEnabled = e.value
-    updateNode({
-      id: id,
-      updatedData: data.internal
-    })
-  }
-
   return (
     <>
       {/* build on top of the Node component */}

renderer/components/learning/results/utilities/parameters.jsx

@@ -15,19 +15,45 @@ import { Column } from "primereact/column"
 const Parameters = ({ params, tableProps, columnNames }) => {
   const [data, setData] = useState([])
   const [selectedRows, setSelectedRows] = useState([])
+
+  const isEmptyOrNull = (value) => {
+    // Check specifically for null or undefined
+    if (value == null) { // Using loose equality (==) checks for both null and undefined
+      return true
+    }
+
+    // Check if the value is an object (but not null, which typeof also calls "object")
+    if (typeof value === 'object') {
+      // A robust check for an empty object: ensure its constructor is Object and it has no own properties
+      return Object.keys(value).length === 0 && value.constructor === Object
+    }
+
+    // Other non-object, non-null values (like strings, numbers, booleans)
+    // are not considered "empty objects" or "null" by this definition.
+    return false
+  }
+
   useEffect(() => {
     if (params) {
       let dataList = []
       Object.keys(params).forEach((key) => {
-        let value = params[key]
-        // For array values
-        if (Array.isArray(value)) {
-          value = JSON.stringify(value)
+        // skip null or undefined values
+        if (isEmptyOrNull(params[key])) {
+          dataList.push({
+            param: key,
+            Value: "null"
+          })
+        } else {
+          let value = params[key]
+          // For array values
+          if (Array.isArray(value)) {
+            value = JSON.stringify(value)
+          }
+          dataList.push({
+            param: key,
+            Value: value != null ? value : "null"
+          })
         }
-        dataList.push({
-          param: key,
-          Value: value != null ? value : "null"
-        })
       })
       setData(dataList)
     }

renderer/components/mainPages/application.jsx

@@ -358,9 +358,8 @@ const ApplicationPage = ({ pageId }) => {
      * @param {Array} columnsArray An array of the columns of the dataset
      */
     const checkWarnings = (columnsArray) => {
-      let datasetColsString = JSON.stringify(columnsArray)
-      let modelColsString = JSON.stringify(modelFeatures)
-      if (datasetColsString !== modelColsString && modelFeatures && columnsArray) {
+      if (!modelFeatures.every((col) => columnsArray.includes(col)) && modelFeatures && columnsArray) {
+        const missingCols = modelFeatures.filter(col => !columnsArray.includes(col))
         setDatasetHasWarning({
           state: true,
           tooltip: (
@@ -371,17 +370,9 @@ const ApplicationPage = ({ pageId }) => {
                 <div style={{ maxHeight: "400px", overflowY: "auto", overflowX: "hidden" }}>
                   <Row>
                     <Col>
-                      <p>Needed columns:</p>
+                      <p>Missing columns:</p>
                       <ul>
-                        {modelFeatures.sort().map((col) => {
-                          return <li key={col}>{col}</li>
-                        })}
-                      </ul>
-                    </Col>
-                    <Col>
-                      <p>Received columns:</p>
-                      <ul>
-                        {columnsArray.sort().map((col) => {
+                        {missingCols.map((col) => {
                           return <li key={col}>{col}</li>
                         })}
                       </ul>

renderer/components/mainPages/superset/SupersetFrame.jsx

@@ -299,7 +299,7 @@ const SupersetDashboard = () => {
         })
       } else if (system === "darwin") {
         // macOS
-        exec(`pkill -f superset`, (err, stdout, stderr) => {
+        exec(`/usr/bin/pkill -f superset`, (err, stdout, stderr) => {
           if (err) {
             console.error(err)
             toast.error("Error killing Superset", {autoClose: 5000})

renderer/styles/flow/reactFlow.css

@@ -121,7 +121,6 @@
 }
 
 .draggable-side-node.p-card {
-  background: #f3f3f3;
   color: #495057;
   box-shadow:
     2px 2px 3px -1px rgb(0 0 0 / 20%),

renderer/styles/theme.css

@@ -15,7 +15,7 @@
   --button-bg: #0d6efd;
   --button-text: #ffffff;
   --button-hover: #0b5ed7;
-  --card-bg: #ffffff;
+  --card-bg: #f8f9fa;
   --card-border: #dee2e6;
   --input-bg: #ffffff;
   --input-border: #ced4da;
@@ -105,7 +105,6 @@ label {
 
 .card-header,
 .p-card-header {
-  background-color: var(--card-bg) !important;
   border-bottom-color: var(--border-color) !important;
   color: var(--text-primary) !important;
 }