Educational Process Analysis

The goal of educational process analysis is to uncover hidden learning and teaching processes in temporal educational data. The process data often include information on what students and teachers do throughout time. We can deduce multiple representations of the unobserved learning and teaching processes that may be generating the observable data from such data. To put it another way, the process data is the product of a collection of hidden learning and teaching processes that take place within students and teachers.

Check out our article in the Journal of Educational Data Mining to learn more about our Process Mining work: EDUCATIONAL PROCESS ANALYSIS MODELING NAEP TEST-TAKING BEHAVIOR

Data-hidden processes can be described by a variety of constructions and models. Directed graphs, process model representations like Petri Nets, Heuristic Nets, and Fuzzy Nets, graphical models like Hidden Markov Models or Bayesian Networks, simpler constructs like Markov Chain Transition Matrices, and even trivial representations like discrete event sequences are just a few examples. Association Rule Mining (Garca et al., 2010), Sequential Pattern Mining (Zhou et al., 2010), Process Mining (Trcka et al., 2010; Bogarn et al., 2018), Graph-Based Analysis (Lynch et al., 2017; Patel et al., 2017), and Curriculum Pacing (Patel et al., 2018) are some of the techniques that can help us discover different representations of The Rule Mining approaches, for example, can be used to determine which student/teacher interactions occur more frequently. Pattern mining techniques can discover common activity sequences in data. Process Models and Graph-Based Analysis can provide useful information.

Data from the educational process can take many different forms and sizes, necessitating the use of various approaches for various sorts of data. For example, graph-based algorithms or process modelling techniques like Heuristic Miner (Bogarn et al., 2018) can be used to assess task-level data with a minimal amount of volatility. When there is a lot of complexity in the data, these techniques become more difficult to employ. This is frequently the case with click-stream data, where we may utilise algorithms like Fuzzy Miner (Bogarn et al., 2018) to provide us more freedom with ‘zooming in and out’ of the process maps, allowing us to easily look at both more and less frequent behaviours. We can utilise sequence clustering algorithms to group the data if there is a lot of variation, which means there are a lot of student learning processes or behaviours that are linked to each other.

Check out another one of our blog posts for a simple example of process mining on educational data: https://www.playpowerlabs.com/post/let-s-do-educational-process-mining

Thank you for taking the time to read this.