Optimal Learning Lab

Applying cognitive theory to optimize learning.
Menu
Reference

Input parameters and output values used in typical LKT workflows.

Reference guide to common input parameters and output values in LKT workflows.

Inputs

Common input parameters

Data and structure

  • data: dataset containing student and outcome information
  • components: factors used to compute features per subject
  • features: methods used to compute those features
  • curvefeats: columns used with difference-style functions
  • connectors: linear-equation operators such as +, *, and :

Fitting and optimization

  • usefolds: folds used for model fitting
  • fixedpars and seedpars: parameter vectors used in fitting and initialization
  • epsilon, cost, type, bias: solver-related settings
  • lowb, highb, maxitv, factrv: optimization controls
  • dualfit, interc, verbose, nosolve: behavior flags
Outputs

Common returned values

Model objects and fit

  • model: the fitted logistic regression model
  • nullmodel: intercept-only baseline model
  • coefs: fitted coefficients
  • r2: pseudo R-squared style fit measure
  • optimizedpars: optimized parameters from fitting

Predictions and derived outputs

  • prediction: predicted probabilities or outcomes
  • latencymodel: latency-related model if dual fitting is enabled
  • studentRMSE: per-student prediction error summary
  • newdata: transformed or feature-enriched data
  • automat: record of automatic transformations or feature engineering