The same engine, many problems | A worked example: fraud detection | The economics | Cost and revenue as we review more transactions | How far down the ranked list should we review? | Choosing a decision threshold | Adapting to your own use case | Regression models: targeting by predicted value | How much value do we capture? | What is the profit of targeting by predicted value?
Package: modelimpact 1.2.0
modelimpact: Assess the Business Impact of Classification Models
Calculate and visualise the financial impact of using a classification model to prioritise or target cases. Although originally designed for customer churn, the same tools apply to many binary classification problems, including fraud detection, credit default, lead scoring and marketing response, upsell and cross-sell, and predictive maintenance. Provides cost, revenue, profit and return-on-investment curves as a function of the share of cases targeted, cumulative gains and lift, marginal profit per bin, and confusion-matrix based payoff across probability thresholds. Also includes 'ggplot2' 'autoplot()' methods and an interactive 'shiny' application for exploring the results.
Authors:
modelimpact_1.2.0.tar.gz
modelimpact_1.2.0.zip(r-4.7)modelimpact_1.2.0.zip(r-4.6)modelimpact_1.2.0.zip(r-4.5)
modelimpact_1.2.0.tgz(r-4.6-any)modelimpact_1.2.0.tgz(r-4.5-any)
modelimpact_1.2.0.tar.gz(r-4.7-any)modelimpact_1.2.0.tar.gz(r-4.6-any)
modelimpact_1.2.0.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
modelimpact/json (API)
| # Install 'modelimpact' in R: |
| install.packages('modelimpact', repos = c('https://peerchristensen.r-universe.dev', 'https://cloud.r-project.org')) |
Bug tracker:https://github.com/peerchristensen/modelimpact/issues
Datasets:
- predictions - Predictions from a customer churn model.
Last updated from:783a51c1dc. Checks:9 OK. Indexed: yes.
| Target | Result | Time | Files | Syslog |
|---|---|---|---|---|
| linux-devel-x86_64 | OK | 143 | ||
| source / vignettes | OK | 197 | ||
| linux-release-x86_64 | OK | 138 | ||
| macos-release-arm64 | OK | 98 | ||
| macos-oldrel-arm64 | OK | 122 | ||
| windows-devel | OK | 91 | ||
| windows-release | OK | 92 | ||
| windows-oldrel | OK | 92 | ||
| wasm-release | OK | 121 |
Exports:bootstrap_profitbreak_evenbudget_frontierbudget_profitcompare_modelsconfusion_payoffcost_revenuecumulative_gainsglanceimpact_reportimpact_summarylift_curvemarginal_profitpayoff_gridplot_budgetplot_cost_revenueplot_gainsplot_liftplot_marginalplot_profitplot_qiniplot_rocplot_roiplot_thresholdsplot_value_gainsprofitprofit_thresholdsqini_curveroc_prroirun_apptidytornadouplift_profitvalue_gainsvalue_profit
Dependencies:clidplyrgenericsgluelifecyclemagrittrpillarpkgconfigR6rlangtibbletidyselectutf8vctrswithr
Readme and manuals
Help Manual
| Help page | Topics |
|---|---|
| Bootstrap confidence bands for the profit curve | bootstrap_profit |
| Break-even and optimal targeting point | break_even |
| Budget-constrained targeting | budget_frontier budget_profit |
| Compare several models on a business-impact metric | compare_models |
| Confusion matrix and payoff at a single threshold | confusion_payoff |
| Calculate cost and revenue | cost_revenue |
| Cumulative gains | cumulative_gains |
| Render a model-impact report | impact_report |
| One-line business-impact summary | impact_summary |
| Lift | lift_curve |
| Marginal profit per bin | marginal_profit |
| Plot modelimpact results | autoplot.mi_bootstrap autoplot.mi_budget autoplot.mi_compare autoplot.mi_cost_revenue autoplot.mi_gains autoplot.mi_lift autoplot.mi_marginal autoplot.mi_profit autoplot.mi_qini autoplot.mi_roc autoplot.mi_roi autoplot.mi_thresholds autoplot.mi_value_gains modelimpact-plots plot_budget plot_cost_revenue plot_gains plot_lift plot_marginal plot_profit plot_qini plot_roc plot_roi plot_thresholds plot_value_gains |
| Tidy and glance methods for modelimpact results | glance.mi_budget glance.mi_profit glance.mi_qini glance.mi_thresholds glance.mi_value_gains modelimpact-tidiers tidy.mi_bootstrap tidy.mi_budget tidy.mi_compare tidy.mi_cost_revenue tidy.mi_gains tidy.mi_lift tidy.mi_marginal tidy.mi_profit tidy.mi_qini tidy.mi_roc tidy.mi_roi tidy.mi_thresholds tidy.mi_value_gains |
| Payoff sensitivity grid | payoff_grid |
| Predictions from a customer churn model. | predictions |
| Calculate profit | profit |
| Find optimal threshold for churn prediction (class) | profit_thresholds |
| Qini (uplift) curve | qini_curve |
| ROC and precision-recall curve data | roc_pr |
| Calculate Return on investment (ROI) | roi |
| Launch the interactive modelimpact Shiny app | run_app |
| Tornado sensitivity analysis of maximum profit | tornado |
| Profit of an uplift campaign | uplift_profit |
| Value gains for a regression model | value_gains |
| Profit for a regression model | value_profit |
