Jonathan Allen founded Allen Analytics LLC in 2009 to provide data-driven discovery in chemical and environmental research.  We offer consulting services at the intersection of our technical expertise and domain knowledge.  Our areas of expertise are chemical engineering, analytical chemistry, statistics, and mathematical modeling.  We have experience in air quality control from coal-fired power plants, and atmospheric aerosols.  Our projects leverage large data sets to address technical issues that affect our client’s bottom line.

A number of our projects have used retrospective data analysis to understand the performance of air quality control systems.  Projects have focused on the performance of pilot systems, newly installed equipment, and systems subject to new regulations.  Our clients posed technical queries with clear connections to their business goals.  We worked with our clients to identify relevant existing data.

Our approach is to first retrieve existing data relevant to the project.  We then analyze data in stages; visualizing and refining the data at each stage.  Using our engineering expertise, we then catalog observations which qualitatively address the client queries; these may be long-term trends or short-term events.  We quantify these observations using statistical tests and semi-empirical engineering models based on the available data.  Throughout the project we regularly report progress and reframe the client queries in light of our findings.  Our goal is to extract knowledge relevant to our client’s business goals.

Our project deliverables include a final narrative report which addresses our client’s queries and business goals.  We also identify opportunities to further leverage our client’s expertise and data; these include suggested process improvements gleaned from anomalous process events.  We also provide data visualizations, the complete data set, and source code of the data analysis software.  All stages of our data analysis including the statistical and engineering models are programmed using the Matlab environment.  The programming approach is self-documenting and scales well for large data sets.