International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 13 Issue: 08 | Aug 2026
p-ISSN: 2395-0072
www.irjet.net
An Integrated Geo-Temporal System for Forecasting Cyber Attacks Using Machine Learning S.V.S. Santhi1, N. Hima Chakradhar2, M. Satwik3, P. Bipin Satyankar4 1,2,3,4Dept. of Computer Science and Engineering (Data Science), Anil Neerukonda Institute of Technology and
Sciences, Visakhapatnam, Andhra Pradesh, India ------------------------------------------------------------------------------***-------------------------------------------------------------------------data to reveal hidden time-based patterns that link to Abstract - In today's digital era, the increasing volume attack incidents and forecast future threats [3]. and sophistication of cyber threats demands intelligent and Organizations can use predictive analytics to shift from automated defences. Machine learning techniques are used defensive security methods to proactive protection to detect, analyse, and predict cyber-attacks, enabling systems because it helps them detect threats early, proactive cybersecurity measures and timely threat distribute resources effectively, and prepare for incidents mitigation. Existing models analyse location and time of more efficiently. cyber-attacks in isolation, without combining them effectively or providing accurate forecasting before the Ensemble learning methods provide superior performance threat occurs. The proposed model overcomes these for modelling complex non-linear data patterns which are limitations by integrating location and time-based features typical in cybersecurity datasets according to machine in a unified framework. The system uses encoded regional learning research. Random Forest algorithms achieve information along with time patterns to understand attack strong predictive accuracy through their method of behaviour. A multi-output regression model is trained to combining multiple decision trees while using bootstrap predict the different attack types and network protocols for aggregation to decrease prediction uncertainty. The two weeks and identifies the most dominant type of attack Random Forest model exhibits effective prediction likely to occur; model performance is evaluated using R², performance across various types of cybersecurity data MAE and RMSE. Hence, the proposed work for forecasting which comprises attack frequency patterns, network cyber-attacks proves to be novel by leveraging geo-temporal abnormality detection, and state-based incident data in a unified and predictive framework. distribution. Key Words: Cyber Attack Prediction, Random Forest, Time Series Forecasting, Multi-Output Regression, Predictive Modeling, Threat Intelligence, GeoTemporal Analysis, Machine Learning.
Predictive analytics is increasingly becoming significant within the context of cybersecurity, as shown by various scholarly findings. According to recent research on intrusion detection systems based on machine learning, the use of ensemble algorithms proves to be efficient in detecting complicated attack patterns [5]. Recent research on the analysis of cyber threat intelligence indicates that predictive modelling should be recognized as the primary method to increase the capacity of national cybersecurity [6]. Time-series models, along with predictive analytics, have developed systems capable of predicting future attacks and measuring threats in various regions [7].
1. INTRODUCTION The growing use of digital technology in government services, financial systems, healthcare systems, and business networks has created a larger cyberspace threat environment. The increasing dependence of countries on their digital networks has led to more complex cyberattacks which include Distributed Denial of Service (DDoS) attacks, malware, ransomware, and network hacking. The Indian government has made protecting essential cyber systems a national priority because the country is expanding its digital transformation projects through cloud technology and e-government services. Traditional cybersecurity mechanisms depend on reactive defence models which only detect and deal with threats after attackers have already entered the system.
The main contributions of this study are briefly listed below: 1) A predictive cybersecurity framework that is aimed at predicting the distribution of cyber-attacks on Indian states through machine learning algorithms. 2) A Random Forest-based multi-target regression framework that can be used to predict both time and spatial cyber-attack patterns.
The latest developments in data science and machine learning technology create fresh possibilities for predictive cybersecurity analytics through their recent advancements. Machine learning models use historical cyber incident datasets together with network telemetry
© 2026, IRJET
|
Impact Factor value: 8.315
3) A data processing, feature extraction, forecasting, and evaluation pipeline for analysing cyber-attacks using machine learning forecasting approaches.
|
ISO 9001:2008 Certified Journal
|
Page 248