pmsims

• Routed calibration-slope searches through CSSE for the machine-learning models. The search looks for the sample size at which performance *equals* the target, which requires the metric to move monotonically with n. The calibration slope does so for classical regression models, which overfit and approach one from below, but not for penalised and tree-based learners, whose slope can approach one from either side depending on how much the fit is shrunk, and so can meet the target at more than one sample size. CSSE is a squared distance from perfect calibration, so it improves monotonically regardless of the direction of miscalibration. Requesting metric = "calibration_slope" with lasso, ridge, rf, or xgboost now searches on the CSSE scale internally and translates the answer back, so targets and results stay on the calibration slope scale.

Sample-size search and results

• Improved adaptive bounds and detection of performance targets that appear unreachable.
• Improved fallback behaviour when model fitting or metric calculation fails.
• Forwarded additional engine arguments consistently through simulate_custom() and the wrapper functions.
• Expanded and standardised the metadata returned by wrapper simulations, which now records the full data-generating configuration alongside the search settings.
• Added an up-front runtime estimate: the adaptive first stage is timed and extrapolated to the full search. Runs expected to take more than 30 minutes report the estimate; runs expected to take more than two hours also warn and ask for confirmation before continuing. The prompt is interactive-only and can be disabled with options(pmsims.confirm_long_runs = FALSE).

Printed output

• Reorganised the console summary into sections: Inputs is now divided into Data-generating scenario, Model and performance, and Simulation, and Results leads with the minimum sample size, followed by the performance expected at that sample size.
• Reported the data-generating configuration under Data-generating scenario: signal complexity, nonlinear strength, predictor distribution, predictor correlation, and binary predictor prevalence.
• Reported performance estimates beneath a Performance at N = ... heading, with the achieved value for the chosen metric shown alongside its target.
• Expressed the sample-size target as a criterion including its direction, for example Calibration slope >= 0.900, using the direction in which the metric improves.
• Replaced internal identifiers with human-readable model names (Random forest rather than rf) and standardised names for statistical quantities (AUC and the C-statistic are both reported as C-statistic for a binary outcome).
• Combined the signal and noise predictor counts onto one line, and stopped repeating the model and mode in Results.
• Moved the mean/assurance explanation and the CSSE footnote below a closing rule, and de-emphasised them.
• Added print(x, verbose = TRUE) for implementation-level detail: internal metric identifiers, the engine settings used for the search, and quantities recorded on an internal search scale. summary() now prints this detailed display.

Interface changes

• Updated the wrapper interface to express predictor configuration through complexity and data_control. The top-level predictor_type and binary_predictor_prevalence arguments have been removed; both are now data_control fields. See the current examples and function documentation for details.
• Candidate predictors are now correlated by default (data_control$correlation = 0.3), whereas 0.5.0 generated independent predictors. Set correlation = 0 to reproduce the earlier behaviour.