Use of statistics and data-analysis methods is a practical necessity for conducting research into virtually any aspect of earth science. As a consequence, instruction in the methods and procedures used most often by earth scientists needs to be provided to students who aspire to careers in this field. In the past, this training has often been organized on an ad hoc or “as needed” basis, usually in conjunction with a specific research project, and usually focused only on the procedures employed in that project. This approach is no longer suitable for the training of undergraduate earth science students insofar as it usually (1) fails to provide adequate grounding in the wide range of procedures and methods available to earth scientists, (2) provides inadequate coverage of the particular application histories, assumptions, strengths and weaknesses inherent in the use of these procedures, and/or (3) neglects to provide focused instruction on the manner in which the results of such applications can be used to test earth science hypotheses. By the same token, general service courses offered by departments of statistics or mathematics often fail to include instruction in procedures that take the specific nature of earth science data into consideration, as well as tending to be delivered in instructional styles that many earth science students find difficult to understand and intimidating. In addition, the advent of new and highly sophisticated data-analysis procedures — such as machine learning and artificial intelligence — are widely acknowledged to have the potential to revolutionize all scientists’ abilities to address many complex problems in their fields, yet discussions of these methods are rarely included in basic data analysis courses.
This course will address these needs of earth science students by drawing on the instructor’s 30+ years of using, developing, promoting, writing about and teaching the skill – and the art – of data analysis in earth science contexts. Offering a trifecta that includes a survey of data-analysis theory, non-mathematical descriptions of how various procedures interact with, and reveal, patterns in datasets, and practical experience with the application of selected procedures to various earth science data-analysis problems, this course will take students with limited backgrounds in mathematics and train them to be confident practitioners of quantitative data analysis. Instruction in probability, basic statistical analysis, regression analyses, time-series analysis, directional data analysis, quantitative stratigraphy, eigenvector methods, dimensionality-reduction procedures and discriminant analysis will all be covered. The course program will then go on to survey the procedures and applications of machine-learning methods and artificial intelligence in the earth sciences. Students who work through the course materials diligently and complete the instructional program successfully will be well positioned to select appropriate data-analysis strategies for a wide range of earth science problems, design and carry out their own descriptive and exploratory analyses, and interpret the results of those analyses with confidence. More importantly though, they will understand the power of these methods to make patterns in earth science data that are invisible to the naked eye and/or causal inspection, visible and in so doing to enable the realization of investigations that would be impossible to deliver objectively in any other manner.
Copyright © 2021 Norman MacLeod - All Rights Reserved.