Analysis of Earthquake Forecasting using Machine Learning with Simulated Data and How can a Visit to a Science Center Influence Students’ Perception of Science?

Sammanfattning

This is a study in two parts, a technical and a pedagogical. The technical part is about forecasting of simulated earthquakes through the means of artificial intelligence (AI) and the pedagogical part examines the effect a visit to a science center changes the way high school students conceptualize science. This technical study tries to answer whether or not a small machine learning algorithm could forecast the size and location of a simulated earthquake based on a simplified model in which radon gas is emitted from the epicenter as a function of time before the earthquake. To achieve this, the simulation included a network of radon detectors distributed in the regions surrounding the epicenter. This model combined with the radon transmission model generated a simulated time-dependent radon signal in each direction. This data was then fed to a neural network and since the data was simulated, a large dataset could be used for training the network. The results show that with the simplified model used, a basic artificial neural network could with an error of < 5% forecast the relative position, magnitude and time of an earthquake one day before the earthquake and 10 days before with an error of ∼ 5%. Since the nature of real radon emission and transport is not known in relation to seismic events, this result is just a demonstration showing that a neural network, given a step, linear or exponential correlation between radon and seismic activity, can forecast the position, time and magnitude of an earthquake. Science centers have recently gained a lot of popularity in Sweden with the aspiration to invoke curiosity for science in students while also offering an alternative learning environment. An alternative that have been argued as something positive in many aspects to learning (Sjöberg 2011, Ersen & Muhlis 2015, De-Witt & Hohenstein 2010) but how does it affect the way students conceptualize science? This study examines 27 high school students’ conceptions of science before and after a visit to a science center by interviewing them in groups before and after. The results indicate that students’ scientific literacy increased, as conceptions regarding stereotypes and that science is something distant were less common in the interviews conducted after the visit. Suggesting that their understanding of scientific research increased to a point where the most common stereotypes were not needed in order to describe their thoughts. An extension of this is feeling like science is something real and closer to the students. The students’ interests and motivation did not change notably, many students were however more secure in their perception that doing research is not for them. A concept that managers decide what scientists do appeared during the first set of interviews but was almost nonexistent afterwards, despite the fact that the science center did not challenge this concept. It’s clear that science centers affect the students’ perception of science to something more along reality and is therefore a good tool for teachers to use as alternative learning environments.