ProcData

ProcData provides R package functions for organizing, extracting features from, and modeling timestamped action sequences (process data) from computer-based assessments to analyze problem-solving behavior.

Key Features:

• Data Organization and Inspection: An S3 class 'proc' for structuring process data with extended generic methods such as summary and print for dataset inspection.
• Feature Extraction: Methods that transform irregular response processes into standardized numerical vectors for downstream analysis.
• Neural Network-Based Predictive Modeling: Recurrent neural network (RNN) models to relate response processes to binary or continuous outcomes, provided as wrappers around the 'keras' library.
• Real-World Dataset Inclusion: Bundled response-process dataset from the 2012 Programme for International Student Assessment (PISA) focused on a climate control item.

Scientific Applications:

• Educational Assessment: Analysis of process data to characterize student problem-solving strategies and inform assessment design and instructional research.

Methodology:

Uses an S3 'proc' class with summary/print methods, feature-extraction routines that convert irregular response processes into numerical vectors, and RNN models implemented via wrappers around the 'keras' library.