zenon/netsec-lab
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Save model

Browse Source
This commit is contained in:
Tobias Eidelpes 2021-06-05 18:43:22 +02:00
parent 72554590fb
commit bd9d3b6932
1 changed files with 6 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
competition/random_forest.py
View File

@ -1,7 +1,6 @@
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sn
import pickle
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import classification_report, confusion_matrix
from sklearn.model_selection import train_test_split
@ -21,9 +20,9 @@ df.drop(['flowStartMilliseconds'], 1, inplace=True)
X = np.array(df.drop(columns=['sublabel']))
y = np.array(df['sublabel'])
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, shuffle=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
clf = RandomForestClassifier(n_estimators=50, n_jobs=-1, criterion='gini', random_state=0, class_weight="balanced")
clf = RandomForestClassifier(n_estimators=50, n_jobs=-1, criterion='gini', random_state=0)
clf.fit(X_train, y_train)
accuracy = clf.score(X_test, y_test)
@ -32,19 +31,11 @@ y_pred_train = clf.predict(X_train)
y_pred_test = clf.predict(X_test)
print("\n *************** TRAINING ****************")
cm_train = confusion_matrix(y_train, y_pred_train)
plt.figure(figsize=(10, 7))
sn.heatmap(cm_train, annot=True)
plt.xlabel('Truth')
plt.ylabel('Predicted')
plt.show()
print(cm_train)
print(classification_report(y_train, y_pred_train))
print("\n ************** VALIDATION ***************")
cm_test = confusion_matrix(y_test, y_pred_test)
plt.figure(figsize=(10, 7))
sn.heatmap(cm_test, annot=True)
plt.xlabel('Truth')
plt.ylabel('Predicted')
plt.show()
print(cm_test)
print(classification_report(y_test, y_pred_test))
example_measure = np.array([ip_to_bin('2.1.1.1'), ip_to_bin('2.1.1.2'), 0, 0, 1])
pickle.dump(clf, open('network_traffic_classifier.sav', 'wb'))