Merge remote-tracking branch 'origin/master'

RF/RF.py

+import pandas as pd
+import numpy as np
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.metrics import ConfusionMatrixDisplay, confusion_matrix, classification_report
+from sklearn.pipeline import Pipeline
+from sklearn.preprocessing import StandardScaler
+
+class RFgenerator :
+    def __init__(self, train_path="../cmake-build-debug/Extraction/train_features.csv", test_path="../cmake-build-debug/Extraction/test_features.csv"):
+        self.train_data = self.loadData(train_path)
+        self.test_data = self.loadData(test_path)
+
+        rd = RandomForestClassifier(max_leaf_nodes=None, max_depth=None, n_estimators=100, criterion='gini')
+        self.RandomForestClassifier = Pipeline([("scaler", StandardScaler()), ("rd", rd)])
+        self.RandomForestClassifier.fit(self.train_data["data"], self.train_data["labels"])
+        self.ypred = self.RandomForestClassifier.predict(self.test_data["data"])
+
+        self.cm = confusion_matrix(self.test_data["labels"], self.ypred)
+        self.cr = classification_report(self.test_data["labels"], self.ypred)
+
+        ConfusionMatrixDisplay(self.cm)
+
+        self.strScore()
+
+    def loadData(self, path):
+        """Loads data"""
+        # Loading
+        file = pd.read_csv(path, sep=',')
+        data = dict()
+        data["data"] = np.array(file.values[:, :-2], dtype=float)
+        data["labels"] = file.values[:, -2].reshape((data["data"].shape[0], 1))
+        data["files"] = file.values[:, -1].reshape((data["data"].shape[0], 1))
+
+        return data
+
+    def strScore(self):
+        print("Confusion Matrix : \n", self.cm)
+        print("classification report : \n", self.cr)
+
+    def strParams(self):
+        print("Data mean : " + str(self.RandomForestClassifier["scaler"].mean_))
+        print("Data std : " + str(self.RandomForestClassifier["scaler"].scale_))
+
+    def code_generate(self):
+        with open("./params.h", "w") as svmdoth:
+            svmdoth.write("#ifndef EML_PARAMS_H\n")
+            svmdoth.write("#define EML_PARAMS_H\n")
+            svmdoth.write("\n")
+            svmdoth.write(self.np2cpp(self.RandomForestClassifier["rd"].classes_, "classes", "std::string") + ";\n")
+            svmdoth.write(self.np2cpp(self.RandomForestClassifier["scaler"].mean_, "mean") + ";\n")
+            svmdoth.write(self.np2cpp(self.RandomForestClassifier["scaler"].scale_, "sigma") + ";\n")
+            svmdoth.write(self.np2cpp(self.RandomForestClassifier["rd"].coef_, "rd_coefs") + ";\n")
+            svmdoth.write(self.np2cpp(self.RandomForestClassifier["rd"].intercept_, "rd_intercept") + ";\n")
+            svmdoth.write("\n")
+            svmdoth.write("#endif //EML_PARAMS_H")
+
+    def np2cppRec(self, x):
+        """Recursive function to convert np array in C++. This recusrive function deals with data of the np array. The header is generated by np2cpp function"""
+        ret = "{"
+        for i in range(x.shape[0]):
+            if type(x[i]) == np.ndarray:
+                if i == x.shape[0] - 1:
+                    ret += self.np2cppRec(x[i])[:-1]
+                else:
+                    ret += self.np2cppRec(x[i])
+            else:
+                ret += str(x[i]) + ", "
+
+        ret = ret[:-2] + "}, "
+
+        return ret + "\n"
+
+    def np2cpp(self, x, name, type="real"):
+        """Generates cpp code associated to a numpy array, no matter its dimensions. It generates the header of the c++ vector. np2cppRec deals with inside data."""
+        ret = "std::"
+        ret += "vector" + (x.ndim-1) * "<std::vector" + "<" + type + ">" + (x.ndim-1) * ">" + " " + name
+
+        ret += self.np2cppRec(x)[:-3]
+
+        return ret
+if __name__ == "__main__":
+    RF = RFgenerator()
+    RF.strParams()
+    RF.code_generate()
\ No newline at end of file