Using Machine Learning to Analyze Detection of Malicious URL

Author:

Andrew Lopez    |       alopez8969@csu.fullerton.edu

Alex Tran       |       quyen137@csu.fullerton.edu

Hyun Woo Kim    |       hyunwoo777@csu.fullerton.edu

Tu Tran         |       trankimtu@csu.fullerton.edu

Summary & Description:

Implemented 4 data modules: 


    - LGC: Linear Regression w/ Count Vectorizer

    - LGT: Linear Regression w/ TFIDF Vectorizer

    - MNBC: Multinomial Naive Bayesian w/ Count Vectorizer

    - MNBT: Multinomial Naive Bayesian w/ TFIDF Vectorizer


Description:


    Cybersecurity has been a great issue worldwide. Through this project, we hope to gain

    insights of how important cybersecurity is and get hands on experience into machine 
    
    learning  technologies to also create a URL scanner for malicious urls to prevent 
    
    exploitation and hacking.


How to compile:

    (Your python 3 interpreter: 3.6 or 3.7 recommended. Not compatible with 3.8)

    python3 ./maliciousURL.py 

        (This allows the user to check options)

    python3 ./maliciousURL.py -t <type> -u <url>
        
        (-t: LGC, LGT, MNBC, MNBT)

        (-u: www.yourwebsite.com/anything-you-want)

Function

    choices: parse the user input as "flags" to connect with other functions

    csvImport: takes in user input url and parse the csv files

    train_test: split the traning and testing into percentile data

    train_test_graph: uses matplotlib.pylot to spit out our dataframe: testing and training of good and bad url

    tokenizerURL: tokenize our input url into meaningful tokens

    vectorizer: vectorize our data from countVectorizer(word count) and  tfidfVectorizer(frequency)

    algorithmReport: generate a report of our data module to Linear Regression and Multinomail Naive Bayes

    LogiRegTFIDF: logistic Regression module with TFIDF vectorizer, attempts to predict our data

    LogRegression_CountVector: logistic Regression module with count vectorizer, attempts to predict our data

    mnbtf: multinomial naive bayes module with TFIDF vectorizer, attempst to predict our data

    mbbcv: multinomial naive bayes module with count vectorizer, attempts to predict our data

    infoDisplay: displays user input choices

    main: passes the correct parameters to the correct function/methods