Strategies for Process Investigations



James McLellanDupuis 316james.mclellan@queensu.ca613-533-2785


Mantong WangDUP

Course Description

This is a course about identifying systematic relationships between variables, and selecting and estimating appropriate models for making predictions about process behaviour and developing insights into the process structure. Different data types and data collection approaches are considered, including quantitative and qualitative data. Graphical and quantitative techniques are presented for exploratory data analysis, identifying possible model structures and parameterizations, and assessing the quality of estimated models and predictions. Both linearly parameterized and nonlinear regression models are discussed, and machine learning techniques are introduced. The distinctions between supervised and unsupervised learning in machine learning are made, and are compared to regression estimation. The roles of designed experiments and data analysis procedures in process investigations are discussed. Applications of two-level factorial and fractional factorial designs in screening studies and higher-order experimental designs are examined. The design component of this course is the planning and execution of an experimental investigation, the analysis of the resulting data, and the formulation of recommendations on the basis of those results.

Prerequisites: CHEE 209 – Analysis of Process Data (or equivalent) or permission of the department. Exclusion: STAT 361.

Objectives and Outcomes

CLO1 Assess the existence of systematic relationships between variables using appropriate graphical and quantitative techniques

KB Math (b)
IN (c)
TOO (a), (c)

CLO2 Estimate empirical models between variables using statistical model building and machine learning techniques including multiple linear and non-linear regression KB Math (b)
IN (c)
TOO (a), (c)
CLO3 Assess the quality of estimated models using graphical and quantitative techniques IN (c)
TOO (a), (c)
CLO4 Evaluate and interpret estimated models taking into account sources of uncertainty and variability IN (c)
TOO (a), (c)
CLO5 Screen and prioritize process variables using 2-level factorial designs, and higher-order experimental designs IN (a)
TOO (a), (c)

This course develops the following attributes at the 4th year level:

Knowledge Base (KB): Math (b) Applies numerical and statistical methods to analyze, interpret and model data

Investigation (IN): (a) Design and conduct investigations to test hypotheses, related to complex problems. (c) Synthesize information from investigation, considering sources of uncertainty and limitations to reach substantiated conclusions.

Engineering Tools (TOO): (a) Develop, adapt and/or extend appropriate software, equipment, models, and simulations for a range of engineering activities. (c) Evaluate limitations and errors of instrumentation/measurement techniques/models/ simulations to assess appropriateness of the results.

Relevance to the Program

Course Structure and Activities

3 lecture hours + 1 tutorial hour per week.  Please refer to Solus for time and locations.


Recommended Textbook: Montgomery, D.C., Runger, G.C., and N.F Hubele, Engineering Statistics, Wiley, New York (2020), 5th Edition (Note: Editions 3, 4, and 5 are acceptable)

Other Materials


The following software will be used:

  • JMP (available from the FEAS software distribution page at no charge (university license)). Please see OnQ course pages for details and instructions.
  • JASP (available for free from the University of Amsterdam). Please see OnQ course pages for details and instructions.
  • Matlab (for machine learning in particular). Available on the computer cluster in Dupuis Hall and in the Teaching Studio (Room 213, Beamish-Munro Hall).


  • Datasets will be made available for practicing analysis and model building. Datasets will be posted on the CHEE418 OnQ pages.

Slides and Readings

  • Slides, readings and supplemental materials will be posted on the CHEE418 OnQ pages