lr_combine: Combine Likelihood Ratios from Multiple Sources

In mispitools: Missing Person Identification Tools

Combine Likelihood Ratios from Multiple Sources

Description

Combines (multiplies) likelihood ratios from two independent evidence sources using the Bayesian multiplication principle. This is used to integrate genetic and non-genetic evidence, or multiple non-genetic variables.

Usage

lr_combine(LRdatasim1, LRdatasim2)

Arguments

LRdatasim1	A data.frame with columns Unrelated and Related containing LR values from the first evidence source. Must be a data.frame (use lr_to_dataframe to convert genetic simulation output first).
LRdatasim2	A data.frame with columns Unrelated and Related containing LR values from the second evidence source.

Details

Under the assumption of conditional independence of evidence given each hypothesis, the combined LR is the product of individual LRs:

LR_{combined} = LR_1 \times LR_2

This follows from Bayes' theorem and is valid when the evidence sources are conditionally independent given the hypothesis.

Important: Both inputs must be data.frames with the same structure. If using output from sim_lr_genetic, first convert it using lr_to_dataframe.

Value

A data.frame with columns:

• Unrelated: Product of LR values under H2

• Related: Product of LR values under H1

The number of rows equals the minimum of the input data frames.

References

Marsico FL, et al. (2023). "Likelihood ratios for non-genetic evidence in missing person cases." Forensic Science International: Genetics, 66, 102891. \Sexpr[results=rd]{tools:::Rd_expr_doi("10.1016/j.fsigen.2023.102891")}

See Also

sim_lr_genetic for genetic LR simulations, sim_lr_prelim for non-genetic LR simulations, lr_to_dataframe for converting genetic simulations, plot_lr_distribution for visualizing combined distributions.

Examples

# Simulate LRs from two different variables
lr_sex <- sim_lr_prelim("sex", numsims = 500, seed = 123)
lr_age <- sim_lr_prelim("age", numsims = 500, seed = 456)

# Combine the evidence
lr_combined <- lr_combine(lr_sex, lr_age)
head(lr_combined)

# Compare distributions
summary(log10(lr_sex$Related))
summary(log10(lr_combined$Related))

# Visualize combined distribution
plot_lr_distribution(lr_combined)

# Combining genetic and non-genetic evidence
library(forrel)
x <- linearPed(2)
x <- setMarkers(x, locusAttributes = NorwegianFrequencies[1:5])
x <- profileSim(x, N = 1, ids = 2)

# Simulate genetic LRs and convert to dataframe
lr_genetic <- sim_lr_genetic(x, missing = 5, numsims = 100, seed = 123)
lr_genetic_df <- lr_to_dataframe(lr_genetic)

# Simulate non-genetic LRs
lr_prelim <- sim_lr_prelim("sex", numsims = 100, seed = 123)

# Combine both sources
lr_total <- lr_combine(lr_genetic_df, lr_prelim)

mispitools documentation built on Aug. 26, 2026, 1:08 a.m.