Using R, Risk assessment in Oil&Gas and More about measures of association

Please join us for pizza at 6 pm.

Talk 1. Using R to assess the risk of microbiologically influenced corrosion in the Oil & Gas industry, by Oscar Montoya

Microorganisms can have beneficial or detrimental impacts on a wide range of human activities, from our food and nutrients (outside and inside our bodies) to human-made infrastructure. Among the negative impacts on infrastructure, microbiologically influenced corrosion in upstream and downstream Oil & Gas industry is a very relevant topic, especially for Alberta, given the close relationship that our province has with fossil fuels production. Some microorganisms, referred as sulfate-reducing microorganisms, are capable of using oil organics as a carbon source (cells building-block material) and respire chemical species present in oil/water, leading to the formation of corrosive compounds. When corrosion happens, it can cause oil spills and the interruption of oil and gas production, ultimately translated into billions of dollars in losses and a negative public image for the industry. Using nucleic acid-sequencing technologies and R, it is possible to identify what microorganisms are present in an oil sample and correlate their presence to environmental data (metadata) to generate very informative conclusions for the industry. The results can then be incorporated into the decision-making toolkit by operators as part of corrosion prevention/management in upstream and downstream operations.

He is a data analyst/computatinoal biologist with a M.Sc. in environmental microbiology from the University of Calgary, and over seven years of experience doing what he likes the most in life: Finding patterns and correlations in complex data sets to answer often intricate questions. A biologist by training, Oscar has over four years of data analytics experience programing in R, with skills that include statistical analysis, advance graphics and report generation. Over the last three years, Oscar has been the sessional lecturer for a graduate-level course on R programing language at the University of Calgary, where he is also doing his Ph.D. on microbiologically influenced corrosion of pipelines. Oscar also has experience with the development of habitat exclusion programs of black vultures (Coragyps atratus) in a landfill, and the use of macroinvertebrates as water quality indicators, and fish characterization.

Talk 2. Some Common and Less Common Measures of Correlations and their Implementation in the R Language, by Thomas Speidel

Correlation is not causation; but it surely is a sign (Good, Hardin. Common Errors in Statistics and How to Avoid Them. Wiley; fourth edition 2012). Measures of correlation are common in many scientific fields as an exploratory tool used to quantify the strength of association between pairs of variables. The ubiquitous Pearson’s correlation coefficient is the default correlation measure in most analytical packages. However, there are more flexible measures in order to cover the eventuality of outliers, non-linear relationships, the type of quantitative variables, or other nuances in the data. In this talk, we will illustrate some common and less common measures of correlations and their applications in R. Time permitting, we will touch upon correlation in time series data where common measures applied to independent data are not appropriate in the presence of dependent data.

He is a Canadian statistician and data scientist, working for Suncor Energy. For nearly a decade he has served as an applied cancer Statistician. Passionate about statistical literacy, he has brought his expertise to the Oil and Gas industry helping engineers and other professionals bridge the gap between the data and the information they need. Thomas' expertise are in the area of predictive and explanatory modelling, time to event analysis, reproducibility and visualizations.

This event is supported by the Pacific Institute for the Mathematical Sciences and Cenovus Energy