plot_2way_comparison: plot_2way_comparison
In gloverd2/codeBase: Contains standard functions for Admiral Advance Analytics

Description Usage Arguments Value Examples

plot to inform which of two models is closer to the actuals. This is done by faceting one score by the other and inspecting if the trend is flat

plot_2way_comparison(
  actual,
  incumbent_pred,
  proposed_pred,
  weight = rep(1, length(actual)),
  n_bins = 9,
  incumbent_label = "Incumbent",
  proposed_label = "Proposed",
  min_weight = 0,
  rebase = FALSE
)

`actual`	Array[Numeric] - Values we are aiming to predict.
`incumbent_pred`	Array[Numeric] - Values that we have predicted by the incumbent model. Incumbent is the current model looking to be replaced.
`proposed_pred`	Optional: String - Text to use to label the proposed predictions.
`weight`	Optional: Array[Numeric] - Weighting of predictions. If NULL even weighting is used.
`n_bins`	integerish: number bins to split data into. Works best if n_bins is square number
`incumbent_label`	Optional: String - Text to use to label the incumbent predictions.
`min_weight`	Optional: Numeric -minimum value of weight to be included in actual. Raise to remove volatility in "actuals" seen in plots
`rebase`	Optional: logical - should the scores be rebased to best match the actuals

a list of 2 ggplot objects. Each of these plots shows one score split by the other.

actual <- rnorm(1000)
incumbent_pred <- actual + (rnorm(1000) * 0.1)
proposed_pred <- actual + rnorm(1000)

plot_2way_comparison(actual=actual,
                     incumbent_pred=incumbent_pred,
                     proposed_pred=proposed_pred)