Network Security Breach Prediction Model

Machine Learning model that detects any breach in the network like infiltration, unauthorized access or any other illegal activities.

Comment

Inspiration

Machine Learning is one of the fastest growing technology in the world. If used efficiently , it can be very helpful for coming up with meaningful insights. Today due to more competition, using machine learning is more important for small enterprises. Artificial Intelligence can power cyber security, customer service and more. With the increasing volume of business data being generated by organizations, AI-based tools can be used for advanced statistical analysis. The use of artificial intelligence and machine learning for small businesses is largely enabling them to manage the shortage of skilled resources and manpower. ML helps in extracting meaningful information from huge datasets. Using ML in business can help in enhancing business scalability, improving business operations. So today small enterprises should use ML effectively to grow their business. So I came up with a project that can be helpful for small enterprises.

What it does

Research said that 60% of small companies were closed after being hacked. Today security is a crucial part of any enterprise and security attacks are the most dangerous things that any business can come across. Network Security breach is any incident that results in unauthorized access to computer data, applications, networks or devices. It results in information being accessed without authorization. Each day security companies discover new threats to business and consumer data. Even if your business is less involved in technology, still your partners can expose you to security threats. It can occur due to your own insiders or malicious outsiders, suspicious data, stolen credentials, third party access and more. Due to this breach, the business should face various consequences such as Revenue Loss, Bad Reputation, Legal Consequences and much more. Beyond the financial consequences of a breach, network security is also hugely important for any business because an attack can compromise the trust of your customers. Therefore, it’s of foremost importance to ensure the highest level of security to lower your risks for a data breach and protect all confidential information. So I came up with a model called Machine Learning powered Network Security Breach Prediction model that detects any breach in the network like infiltration, unauthorized access or any other illegal activities. Instead of dealing with threats after losses, this model looks for unusual activity, so they can stop attacks or raise alerts earlier.

How we built it

In the dataset, we have to predict whether the network is normal or anomaly, i.e, chances of being hacked. Initially I have imported all libraries and dataset. Then Exploratory Data Analysis and Data Visualization was performed on data, which can give meaning insights, discover patterns, detect outliers and can make assumptions with help of graphical representations. Followed by Plotting count of important features like protocol type, service, flag, logged_in, class. Finding correlation of all features, which can help to retrieve important features and drop least important features. Converting categorical features to numerical values using Feature Mapping. Splitting data into training and test set. For getting best parameters for the model I used RandomizedSearchCV which is a cross validation method. Various models were built in this project using - XGBoost Classifier Algorithm - Decision Tree Classifier Algorithm - Random Forest Classifier - Gradient Boosting Algorithm - AdaBoosting Algorithm - MLP Classifier Algorithm - KNN algorithm - Naive Bayes Algorithm. After fitting all these model, confusion matrix and accuracy of all the models have been compared and analyzed. From the graph we can say that XGBoost and Random Forest gave almost 99% accuracy followed by other algorithms.

Challenges we ran into

Finding the best features for model building is the main challenge I came across. By using correlation heat map I was able to find the best supporting features.

Accomplishments that we're proud of

I was able to build a good predictive machine learning model which can help the small enterprises to grow their business.

What we learned

Machine Learning, Data Visualization , Data Cleaning

What's next for Network Security Breach Prediction Model

Next I wish to integrate this machine learning model into a web app.

Built With

Share this project:

Updates