St Edward's Academic Review 2025

ST EDWARD’S, OXFORD

occur. Based on this information, scientists project that 44 earthquakes with Richter magnitudes of 7.5 to 7.6 will occur within the next 35 years. Even though some areas are more vulnerable to earthquakes than others, seismologists were unable to forecast their position and timing as of 2010 (Massachusetts Institute of Technology, 2012). According to an article titled ‘Prediction as an Impediment to Preparedness: Lessons from the US Hurricane and Earthquake Research Enterprises’, ‘all earthquake and hurricane research endeavours in the United States have the same goal of increasing the resilience of vulnerable communities, but they use different methods to achieve this goal’ (Maricle, 2011). In the case of earthquakes, conducting research in a collaborative, responsive, or transparent manner promotes both high quality and usable preparedness-focused science. This is in stark contrast to hurricane research, which is presented as not collaborative, responsive, or transparent because it only promotes predictions and assumes that usability will follow. Due to the vast differences in research methods for hurricanes and earthquakes, forecasts tended to fail in the early 2010s because prediction capabilities were ineffective, even though statistical modelling aided in extrapolating patterns from past data (Massachusetts Institute of Technology, 2012). Maricle argues that the difference between the two research methods is that the predictions made for hurricanes are good but not used effectively while predictions made from earthquake research are more accurate since they involve a collaborative process. It can be deduced that in order to make accurate predictions, a collaborative, responsive and transparent research process is needed to make predictions that increase the resilience of communities that are vulnerable to natural disasters. Maricle also suggests that in the early 2010s there were limited resources for natural disaster research. In all, it can be said that statistical regularity does not mean that natural disasters are predictable, possibly due to the uniqueness of each natural disaster. Therefore, it seems that data science has been limited thus far in predicting natural disasters. The extent to which data science has become more effective in predicting natural disasters Over the period between 2010 and 2020, many researchers have found that data science has become

the information may not be the most accurate, it can be used to make comparisons. The source is extremely useful because it leads to further questions about the evolution of data science in the last decade and whether it has become more efficient in finding patterns and making predictions. In recent years researchers have collected data and developed models that predict disasters (Lavrskyi, 2019). According to Lavrskyi’s research published as ‘Data Science Usage in Natural Disasters Predictions’, having more data makes computational models slower and more complex but he suggests that predictive models should work and make corrections in real time. Furthermore, he explains that artificial intelligence can find hidden dependencies in data that can be a basis for better understanding the mechanism of disasters which could help to improve predictions. The source also gives examples of how data science has aided in predicting natural disasters. These include earthquakes, hurricanes and volcanic eruptions among others. This source has been useful because it answers some of my research questions. It has also given more depth to my research and leads to more questions including how to improve computational models to become faster and less complex. In conclusion, there is a broad range of resources relating to this topic. By comparing the older sources to the more recent ones, it is evident that there has been a significant advancement in the technology used to predict natural disasters. However, there is still a long way to go in terms of improving the accuracy of models used in order to make more precise predictions. The extent of the effectiveness of data science in making predictions during the period 2010-2015 In recent times, the world has experienced the deadliest natural disasters in history. Some examples of these are Hurricane Katrina in 2005, the 2003 Kashmir earthquake, the 2004 Indian Ocean Tsunami, and the 2010 Haiti earthquake (Reuters, 2023). According to the Massachusetts Institute of Technology (2010), scientists concluded that it is impossible to predict the timing and location of a large-scale earthquake but the intensity and frequency of earthquakes show consistent patterns. Each year, roughly 134 earthquakes measure 6.0 to 6.9 on the Richter scale, approximately 17 ranging from 7.0 to 7.9, and one measuring 8.0 or higher

52