School of Population Health


Features

SPAN has several features:

  • complexity penalised search partition analysis - searches for optimal (Boolean) prediction and decision rules
  • graphics for exploratory data analysis - bar charts, empirical density estimation, survival plots, scatter diagrams
  • selection of optimal cut-points
  • ROC and QROC plots
  • criteria for partition effectiveness (mean square error, entropy, quality indices, log-rank, chi-square, Gini index)
  • simple classification and regression decision trees, allowing Boolean splits
  • fitting and displaying scaled rectangle diagrams and mosaic plots
  • incidence rate statistics for survival data and Kaplan-Meier survival
  • cross tabulation statistics
  • by-group analysis (allows overlapping groups)
  • allows Cross-validation
  • handles missing data
  • randomisation tests for statistical significance.