Abstract:The paper considers exploratory data analysis and approaches in predictive analytics for air alerts during the Russian-Ukrainian war which broke out on Feb 24, 2022. The results illustrate that alerts in regions correlate with one another and have geospatial patterns which make it feasible to build a predictive model which predicts alerts that are expected to take place in a certain region within a specified time period. The obtained results show that the alert status in a particular region is highly dependable on the features of its adjacent regions. Seasonality features like hours, days of a week and months are also crucial in predicting the target variable. Some regions highly rely on the time feature which equals to a number of days from the initial date of the dataset. From this, we can deduce that the air alert pattern changes throughout the time.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: during the Russia - Ukraine war, how to predict the occurrence time and duration of air - raid alarms through predictive analysis. Specifically, the author hopes that through exploratory data analysis and the establishment of a prediction model, it will be possible to predict in advance whether an air - raid alarm will be issued in a certain area within a specific future time period, as well as the duration of the alarm. ### Problem Background Since Russia invaded Ukraine on February 24, 2022, understanding the structure and pattern of air - raid alarms has become very important. Accurately predicting the time and duration of air - raid alarms can help residents better cope with potential threats, reduce unnecessary panic, and improve the efficiency of emergency response. ### Research Objectives 1. **Exploratory Data Analysis**: Analyze historical data to find out the spatio - temporal distribution patterns of air - raid alarms in different areas. 2. **Build a Prediction Model**: Based on the results of exploratory data analysis, construct a machine - learning model that can predict future air - raid alarms. ### Key Findings - **Geospatial Correlation**: There is a significant correlation between air - raid alarms in different areas, especially a higher correlation between adjacent areas. This means that the alarm information in one area can provide early warnings for other areas. - **Importance of Temporal Features**: Seasonal features (such as hour, day of the week, month) and the number of days since the initial date of the data set (`ndays`) have an important impact on the prediction results. These features indicate that the pattern of air - raid alarms changes over time. - **Influence of Neighboring Areas**: The features of neighboring areas are of higher importance for the prediction of the target area. ### Method Overview - **Data Sources**: Multiple public data sets were used, especially the air - raid alarm data set provided by Vadym Klymenko. - **Feature Engineering**: A feature set including binary features (indicating whether there is an alarm) and the cumulative alarm duration in the area was constructed. - **Model Selection**: The random forest algorithm was adopted as a classification model and verified through the division of the training set and the test set. - **Evaluation Metrics**: Metrics such as the ROC curve and accuracy were used to evaluate the model performance. ### Conclusions The research shows that by combining geospatial correlation and temporal features, the occurrence of air - raid alarms can be effectively predicted. The alarm information in neighboring areas has important reference value for predicting the future alarm situation in a certain area. In addition, as time passes, the pattern of air - raid alarms is constantly changing, so it is very necessary to dynamically adjust the prediction model. Through these studies, the author hopes to provide more accurate air - raid alarm predictions for the residents and relevant departments in Ukraine, thereby improving the ability to respond to emergencies.

Predictive Analytics of Air Alerts in the Russian-Ukrainian War

Spatiotemporal analysis of drone operations using armed conflict location and event data (ACLED): Focusing on the Russia-Ukraine war

Sentiment Analysis and Comprehensive Evaluation of Supervised Machine Learning Models Using Twitter Data on Russia–Ukraine War

Machine learning model to project the impact of Ukraine crisis

Environmental impacts of shifts in surface urban heat island, emissions, and nighttime light during the Russia-Ukraine war in Ukrainian cities

The Use of Remote Sensing Data for Investigation of Environmental Consequences of Russia-Ukraine War

Monitoring of Urban Changes with multi-modal Sentinel 1 and 2 Data in Mariupol, Ukraine, in 2022/23

Spatio-temporal variations of aerosol optical depth over Ukraine under the Russia-Ukraine war

The spatial dynamics of Ukraine air quality impacted by the war and pandemic

Wildfire spreading prediction using multimodal data and deep neural network approach

Big spatio‐temporal data mining for emergency management information systems

Monitoring of Urban Changes With Multimodal Sentinel 1 and 2 Data in Mariupol, Ukraine, in 2022/23

The practice of connectionist model for predicting forest fires in the Arctic zones of the Krasnoyarsk Territory

Implications of Russia–Ukraine war on land surface temperature and air quality: long-term and short-term analysis

Modelling the suitability of multiple launch rocket system in the war in Ukraine

Detecting the Impacts of Ongoing Russia-Ukraine Conflict on Crop Growth and Human Activity Using Multi-Modal Remote Sensing Time-Series Data

An Open-Source Tool for Mapping War Destruction at Scale in Ukraine using Sentinel-1 Time Series

Sentiment analysis for measuring hope and fear from Reddit posts during the 2022 Russo-Ukrainian conflict

Rapid monitoring of cropland primary productivity and shipping activity in Ukraine

War drives forest fire risks and highlights the need for more ecologically-sound forest management in post-war Ukraine

NLP Case Study on Predicting the Before and After of the Ukraine-Russia and Hamas-Israel Conflicts