Mean signed deviation (also known as mean signed difference, or mean signed error) computes the average differences between truth and estimate. A related metric is the mean absolute error (mae()).

## Usage

msd(data, ...)

# S3 method for data.frame
msd(data, truth, estimate, na_rm = TRUE, case_weights = NULL, ...)

msd_vec(truth, estimate, na_rm = TRUE, case_weights = NULL, ...)

## Arguments

data

A data.frame containing the columns specified by the truth and estimate arguments.

...

Not currently used.

truth

The column identifier for the true results (that is numeric). This should be an unquoted column name although this argument is passed by expression and supports quasiquotation (you can unquote column names). For _vec() functions, a numeric vector.

estimate

The column identifier for the predicted results (that is also numeric). As with truth this can be specified different ways but the primary method is to use an unquoted variable name. For _vec() functions, a numeric vector.

na_rm

A logical value indicating whether NA values should be stripped before the computation proceeds.

case_weights

The optional column identifier for case weights. This should be an unquoted column name that evaluates to a numeric column in data. For _vec() functions, a numeric vector, hardhat::importance_weights(), or hardhat::frequency_weights().

## Value

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 msd_vec(), a single numeric value (or NA).

## Details

Mean signed deviation is rarely used, since positive and negative errors cancel each other out. For example, msd_vec(c(100, -100), c(0, 0)) would return a seemingly "perfect" value of 0, even though estimate is wildly different from truth. mae() attempts to remedy this by taking the absolute value of the differences before computing the mean.

This metric is computed as mean(truth - estimate), following the convention that an "error" is computed as observed - predicted. If you expected this metric to be computed as mean(estimate - truth), reverse the sign of the result.

Other numeric metrics: ccc(), huber_loss_pseudo(), huber_loss(), iic(), mae(), mape(), mase(), mpe(), poisson_log_loss(), rmse(), rpd(), rpiq(), rsq_trad(), rsq(), smape()

Other accuracy metrics: ccc(), huber_loss_pseudo(), huber_loss(), iic(), mae(), mape(), mase(), mpe(), poisson_log_loss(), rmse(), smape()

Thomas Bierhance

## Examples

# Supply truth and predictions as bare column names
msd(solubility_test, solubility, prediction)
#> # A tibble: 1 × 3
#>   .metric .estimator .estimate
#>   <chr>   <chr>          <dbl>
#> 1 msd     standard     -0.0143

library(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) %>%
msd(solubility, prediction)

metric_results
#> # A tibble: 10 × 4
#>    resample .metric .estimator .estimate
#>    <chr>    <chr>   <chr>          <dbl>
#>  1 1        msd     standard   -0.0119
#>  2 10       msd     standard   -0.0424
#>  3 2        msd     standard    0.0111
#>  4 3        msd     standard   -0.0906
#>  5 4        msd     standard   -0.0859
#>  6 5        msd     standard   -0.0301
#>  7 6        msd     standard   -0.0132
#>  8 7        msd     standard   -0.00640
#>  9 8        msd     standard   -0.000697
#> 10 9        msd     standard   -0.0399

# Resampled mean estimate
metric_results %>%
summarise(avg_estimate = mean(.estimate))
#> # A tibble: 1 × 1
#>   avg_estimate
#>          <dbl>
#> 1      -0.0310