modelimpact 1.1.0
New features
- New ranking-based views:
cumulative_gains(),
lift_curve(), marginal_profit(),
break_even() and impact_summary().
- New threshold / classification views:
confusion_payoff(), roc_pr() and
payoff_grid().
- New comparison and robustness tools:
compare_models(),
bootstrap_profit() and tornado().
- All result objects now have
ggplot2::autoplot()
methods, plus matching plot_*() helper functions, for
one-line visualisation.
- New
run_app() launches an interactive
shiny application to explore the package and upload your
own data (comma- or semicolon-separated CSV).
profit(), cost_revenue() and
roi() gained a positive argument so the event
class no longer has to be labelled "Yes", and a
prob_accept argument that reconciles them with
profit_thresholds().
Bug fixes
profit_thresholds() now correctly applies the
fp_val (false-positive value) argument, which was
previously ignored.
modelimpact 1.0.0
- Added functionality for finding optimal thresholds