# parameter_optimization_dga_analysis

This repo contains works on classification of malicious and non-malicious URL using Machine Leraning by tuning the n_gram range. To run the code snniptes:

pip install virtualenv
virtualenv mypython
pip install requirements_latest.txt
source mypython/bin/activate

• The dataPreprocessing.zip file contains code for dataset modification
• final__csv.csv contains the malicious and non-malicious URL labeled as 0 and 1
• res.csv contains only benign URL(80 class)
• Modified_final_updated_26_07_2020.ipynb contains all the necessary code for classification
• accuracy_check.ipynb contains the functions for accuracy check called in Modified_final_updated_26_07_2020.ipynb
• cnn_1d.ipynb contains all the necessary code for classification using Convolutional neural network (1D)
• tuning_logisticRegression.ipynb and tuning_svm.ipynb grid search to find the best parameters

Tuning instructions

• In order to check and run the n_gram range and ML Model run Modified_final_updated_26_07_2020.ipynb and change the variable below.
• Example 1:

   n1 = 1 (n1 and n2 for N_gram rannge.e.g (1,1) will be unigram
   n2 = 1
   model_name = "LinearSVC"

• Example 2

   n1 = 2
   n2 = 2
   model_name = "MultinomialNB"

'''

 For CNN_1D run the cnn_1d.ipynb