README.md

In lucas-johnson/agreeable: Assess Agreement of Datasets

agreeable

The goal of agreeable is to provide a set of simple tools to assess agreement between two matching datasets. The specific use case under which this package was developed is to assess the agreement between modeled and observed datasets.

Computations following Ji and Gallo 2006.

Examples

Here are some of the basic tools made available by agreeable…

Simulated data generation:

library(agreeable)
library(ggplot2)
library(knitr)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(data.table)
#> 
#> Attaching package: 'data.table'
#> The following objects are masked from 'package:dplyr':
#> 
#>     between, first, last

set.seed(123)
data <- data.table(
    observed = sort(rnorm(500, mean = 300, sd = 75)),
    predicted_systematic = sort(rnorm(500, mean = 275, sd = 100))
)
sds <- c(1, 5, 10, 30, 50, 75, 100, 150)
data$predicted_unsystematic <- sort(rnorm(500, mean = 300, sd = sds))

1. AC

Compute AC (agreement coefficient), ACs (systematic agreement), ACu (unsystematic agreement), PUD (percentage of unsystematic difference), and PSD (percentage of systematic difference):

Systematic differences dominate:

knitr::kable(
    data.table::data.table(
        AC = data %>% ac(observed, predicted_systematic),
        ACs = data %>% acs(observed, predicted_systematic),
        ACu = data %>% acu(observed, predicted_systematic),
        PUD = data %>% pud(observed, predicted_systematic),
        PSD = data %>% psd(observed, predicted_systematic)
    )
)

| AC | ACs | ACu | PUD | PSD | |----------:|----------:|----------:|----------:|----------:| | 0.8659736 | 0.8691102 | 0.9968634 | 0.0234028 | 0.9765972 |

Unsystematic (random) differences dominate:

knitr::kable(
    data.table::data.table(
        AC = data %>% ac(observed, predicted_unsystematic),
        ACs = data %>% acs(observed, predicted_unsystematic),
        ACu = data %>% acu(observed, predicted_unsystematic),
        PUD = data %>% pud(observed, predicted_unsystematic),
        PSD = data %>% psd(observed, predicted_unsystematic)
    )
)

| AC | ACs | ACu | PUD | PSD | |----------:|----------:|---------:|----------:|----------:| | 0.8553422 | 0.9967902 | 0.858552 | 0.9778108 | 0.0221892 |

2. GMFR

Scatter plot with 1:1 and Geometric Mean Functional Relationship (GMFR) lines:

Systematic differences dominate:

inter <- data %>% gmfr_intercept(observed, predicted_systematic)
slope <- data %>% gmfr_slope(observed, predicted_systematic)

ggplot2::ggplot(data, aes(x=observed, y = predicted_systematic)) +
    ggplot2::geom_point(
        alpha = 0.7, 
        color = "black"
    ) + 
    ggplot2::geom_abline(aes(slope = 1, intercept = 0, color = "black")) + 
    ggplot2::geom_abline(aes(slope = slope, intercept = inter, color = "orange")) + 
    ggplot2::scale_color_identity(
        labels=c("1:1", "GMFR"), 
        guide="legend", 
        name = ""
    )

Unsystematic (random) differences dominate:

inter <- data %>% gmfr_intercept(observed, predicted_unsystematic)
slope <- data %>% gmfr_slope(observed, predicted_unsystematic)

ggplot2::ggplot(data, aes(x = observed, y = predicted_unsystematic)) + 
    ggplot2::geom_point(
        alpha = 0.7, 
        color = "black"
    ) + 
    ggplot2::geom_abline(aes(slope = 1, intercept = 0, color = "black")) + 
    ggplot2::geom_abline(aes(slope = slope, intercept = inter, color = "orange")) + 
    ggplot2::scale_color_identity(
        labels=c("1:1", "GMFR"), 
        guide="legend", 
        name = ""
    )

Installation

There is no released version of agreeable yet.

You can install the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("lucas-johnson/agreeable")

lucas-johnson/agreeable documentation built on Dec. 21, 2021, 11:52 a.m.