Imputation of missing data in life‐history trait datasets: which approach performs the best?

By C. Penone, Ana D. Davidson, K. Shoemaker, Moreno Di Marco, C. Rondinini, T. Brooks, Bruce E. Young, Catherine H. Graham, G. C. Costa. 2014

396 citations recorded by the source corpus.

Despite efforts in data collection, missing values are commonplace in life‐history trait databases. Because these values typically are not missing randomly, the common practice of removing missing data not only reduces sample size, but also introduces bias that can lead to incorrect conclusions. Imputing missing values is a potential solution to this problem. Here, we evaluate the performance of four approaches for estimating missing values in trait databases (K‐nearest neighbour (kNN), multivariate imputation by chained equations (mice), missForest and Phylopars), and test whether imputed datasets retain underlying allometric relationships among traits. Starting with a nearly complete trait dataset on the mammalian order Carnivora (using four traits), we artificially removed values so that the percent of missing values ranged from 10% to 80%. Using the original values as a reference, we assessed imputation performance using normalized root mean squared error. We also evaluated whether including phylogenetic information improved imputation performance in kNN, mice, and missForest (it is a required input in Phylopars). Finally, we evaluated the extent to which the allometric relationship between two traits (body mass and longevity) was conserved for imputed datasets by looking at the difference (bias) between the slope of the original and the imputed datasets or datasets with missing values removed. Three of the tested approaches (mice, missForest and Phylopars), resulted in qualitatively equivalent imputation performance, and all had significantly lower errors than kNN. Adding phylogenetic information into the imputation algorithms improved estimation of missing values for all tested traits. The allometric relationship between body mass and longevity was conserved when up to 60% of data were missing, either with or without phylogenetic information, depending on the approach. This relationship was less biased in imputed datasets compared to datasets with missing values removed, especially when more than 30% of values were missing. Imputations provide valuable alternatives to removing missing observations in trait databases as they produce low errors and retain relationships among traits. Although we must continue to prioritize data collection on species traits, imputations can provide a valuable solution for conducting macroecological and evolutionary studies using life‐history trait databases.

Subfields: Ecology, Genetics, Paleontology

View Paper: https://www.semanticscholar.org/paper/818c5feea02f6284e9e3efa0a5fbcc0ba600504b

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Ecology

In Genetics

In Paleontology

Citations in the corpus

Hot and bothered: Using trait-based approaches to assess climate change vulnerability in reptiles (2016, background): “…2) making resulting species trait databases openly available, thus giving transparency to the process, enhancing scrutiny of the data and encouraging input from species experts, and 3) define best practice for collecting missing trait data to reduce uncertainty in assessments (Penone et al., 2014).”
Decoupling phylogenetic and functional diversity to reveal hidden signals in community assembly (2017, positive): “An approach to account for missing or complex trait information is to use phylogenetic differences between species (Connolly et al. 2011; Cadotte, Dinnage & Tilman 2012; Penone et al. 2014; Lososov a et al. 2015).”
Comparison of Performance of Data Imputation Methods for Numeric Dataset (2019, methodology): “The study conducted by Penone et al. (2014) evaluates performance of four approaches, for estimating missing values in traits database, namely kNN, mice, missForest, and Phylopars.”
Global gaps in trait data for terrestrial vertebrates (2020, methodology): “Penone et al. (2014) introduced missing values (10%–80”
Climate change vulnerability assessment of species (2018, background): “Progress towards imputing unknown trait data also helps to fill data gaps (Penone et al., 2014; Schrodt et al., 2015).”
Handling missing values in trait data (2020, methodology): “Imputation has been suggested to improve when phylogenetic information is incorporated (Kim et al., 2018; Penone et al., 2014).”
Machine learning and deep learning—A review for ecologists (2022, neutral): “The NLP community demonstrate how this could be done: Model hubs with many different pre-trained models and a simple and common interface that can be used by everyone (e.g. Mäder et al., 2021; Wolf et al., 2019; cf. Ott & Lautenschlager, 2021 ).”