1. Temporal datasets
In this datasets students were able to answer the same question several times.
**Assistments 2009-2010**
**Berkeley** -1730  students tried 234 questions that were from online CS course. 

2. Non-temporal datasets
In this case all students tried all questions.
**Castor** and **Fraction** - middle school students datasets.
**TIMSS** - 757 students and 23 math questions.
**ECPE** -  2922 students and 28 language related tasks.

San Diego State University

Experiments (Knowledge Tracing Machines: Factorization Machines for Knowledge Tracing)

Datasets (Knowledge Tracing Machines: Factorization Machines for Knowledge Tracing)

Based on student ids items and outcomes the authors generate the number of successful/unsuccessful attempts. They use 5-fold cross validation. and divide data into train and test sets. Non temporal datasets are trained for 1000 epochs , assistments - 500 and Berkeley - 300.

Learn Before

Related