Prediction of Malwares in Microsoft Windows Operating Systems

Malware has been identified as one of the predominant cyber threats with the fast growth of the internet. Anti-malware vendors nowadays receive a huge amount of speculated malware files daily. To keep up with the flow of these malware-ridden files, machine-learning techniques are used to abstract similar malwares. This paper presents a prediction model for the prediction of malware attacks based on certain features of the machine and the Light Gradient Boost Algorithm was employed for this purpose. A dataset that came in two splits (the train and test splits) with entries of Windows machines with different specifications was acquired and preprocessed to remove irrelevant features. The train split of the dataset was then used to train the Light Gradient Boost Algorithm to derive a model which was then used for the prediction on the test split. The accuracy of the model was found to be 98% while the precision and recall of the model were also found to be 98%. This study would help Windows users know what kind of specifications of machines are more prone to malware attacks.