abhimanyu-atri / phishing-website-classifier Goto Github PK

An machine learning model that classifies a website as phising website or not by using various classification algorithm and comparing their results.

Following libraries are used in this whole project

• numpy
• pandas
• seaborn
• sciket-learn
• matplotlib
• Pickle

• Pandas is used to load the dataset.
• Separation of features from target variable.
• Seborn is used to display the heatmap of features.
• Division of the dataset into 80-20 for training and testing.
• Then KNN algorithm is applied by using hyperparameter tuning for best results.
• For hyperparameter tuning GridSearchCV is used.
• Then accuracy and confusion matrix is calculated.
• Similarly Naive Bayes, Support vector machine(SVM), decsion tree and Random forest is applied.
• Then pickle dump is used to store the accuracy to the file.

Following are the result of hyperparameter tuning and algoritms applied.

• K Nearest Neighbour

  

• Naive bayes

  

• Support Vector Machine

  

• Decision Tree

  

• Random Forest

  

jupyter nbconvert --ClearOutputPreprocessor.enabled=True --inplace notebooks/*.ipynb