Incident Classification, Prediction of Location and Casualties

Abstract

Indian Army is always being troubled by the constant attacks of the militants and different terrorist organizations. Such encounters always cost us the lives of the soldiers who work for us with an immeasurable selfless attitude, many a times getting slaughtered in different regions of India. Such brainwashed mobs exploit the loopholes of the system, where the ground level officials don't have on-the-spot decision making authority. They use guns, hand grenades, Improvised Explosive Devices (IED) as weapons. Incidents of stone pelting, damage to property, fire incident, mob lynching, etc set the tone for an unrest society. Unfortunately to deal with the rage of such brainwashed minds, soldiers are the first line of defense and are put into a heated situation which leads to bloodshed. To avoid this, authors implemented a model which uses Artificial Intelligence, to predict the location of the next probable ‘Anti-National’ incident and the number of casualties in that incident. By using the principles of Timeseries Analysis, Auto Regressive Integrated Moving Average model in combination with Random Forest Regressor, authors predict the location of the next probable incident and the number of casualties in it. Using Beautiful Soup Library database is created by scraping the news from webpages. A Logistic Regression, is used to classify whether the following news is ‘Anti-National’ or not. The developed algorithm is used effectively to find i) the patterns in these attacks, ii) the factors which spark such incidents and help to take precautionary actions or completely avoid them. This model implementation can track down the activities and assist the Indian Army.