Calculate the index of ideality of correlation. This metric has been studied in QSPR/QSAR models as a good criterion for the predictive potential of these models. It is highly dependent on the correlation coefficient as well as the mean absolute error.

Note the application of IIC is useless under two conditions:

When the negative mean absolute error and positive mean absolute error are both zero.

When the outliers are symmetric. Since outliers are context dependent, please use your own checks to validate whether this restriction holds and whether the resulting IIC has interpretative value.

The IIC is seen as an alternative to the traditional correlation coefficient and is in the same units as the original data.

iic(data, ...) # S3 method for data.frame iic(data, truth, estimate, na_rm = TRUE, ...) iic_vec(truth, estimate, na_rm = TRUE, ...)

data | A |
---|---|

... | Not currently used. |

truth | The column identifier for the true results
(that is |

estimate | The column identifier for the predicted
results (that is also |

na_rm | A |

A `tibble`

with columns `.metric`

, `.estimator`

,
and `.estimate`

and 1 row of values.

For grouped data frames, the number of rows returned will be the same as the number of groups.

For `iic_vec()`

, a single `numeric`

value (or `NA`

).

Toropova, A. and Toropov, A. (2017). "The index of ideality
of correlation. A criterion of predictability of QSAR models for skin
permeability?" *Science of the Total Environment*. 586: 466-472.

Other numeric metrics:
`ccc()`

,
`huber_loss_pseudo()`

,
`huber_loss()`

,
`mae()`

,
`mape()`

,
`mase()`

,
`mpe()`

,
`rmse()`

,
`rpd()`

,
`rpiq()`

,
`rsq_trad()`

,
`rsq()`

,
`smape()`

Other accuracy metrics:
`ccc()`

,
`huber_loss_pseudo()`

,
`huber_loss()`

,
`mae()`

,
`mape()`

,
`mase()`

,
`mpe()`

,
`rmse()`

,
`smape()`

Joyce Cahoon

# Supply truth and predictions as bare column names iic(solubility_test, solubility, prediction)#> # A tibble: 1 x 3 #> .metric .estimator .estimate #> <chr> <chr> <dbl> #> 1 iic standard 0.890library(dplyr) set.seed(1234) size <- 100 times <- 10 # create 10 resamples solubility_resampled <- bind_rows( replicate( n = times, expr = sample_n(solubility_test, size, replace = TRUE), simplify = FALSE ), .id = "resample" ) # Compute the metric by group metric_results <- solubility_resampled %>% group_by(resample) %>% iic(solubility, prediction) metric_results#> # A tibble: 10 x 4 #> resample .metric .estimator .estimate #> <chr> <chr> <chr> <dbl> #> 1 1 iic standard 0.730 #> 2 10 iic standard 0.731 #> 3 2 iic standard 0.906 #> 4 3 iic standard 0.877 #> 5 4 iic standard 0.732 #> 6 5 iic standard 0.821 #> 7 6 iic standard 0.896 #> 8 7 iic standard 0.867 #> 9 8 iic standard 0.881 #> 10 9 iic standard 0.748#> # A tibble: 1 x 1 #> avg_estimate #> <dbl> #> 1 0.819