cito
Deep neural networks in R with ordinary formula syntax, plus the explainable-AI tools to interpret them.
Downloads per year
43,197 downloads from the Posit (RStudio) CRAN mirror since 2022, through 28 October 2026
Cited 9 times on Google Scholar (as of 29 October 2026)
cito fits and interprets deep neural networks in R. A network is
specified with the same formula syntax as lm or glm, hyperparameters can
be tuned by cross-validation in the same call, and the package helps to detect
and handle convergence problems.
Predictions alone are rarely enough in ecology, so cito also provides the tools to look inside a fitted network: variable importance, partial dependence and accumulated local effect plots, and effect estimates, each optionally with confidence intervals and p-values. It is built on the R package torch, needs no Python installation, and runs on CPUs, GPUs and Mac GPUs.
Authors: Christian Amesöder, Maximilian Pichler (maintainer). Contributors: Florian Hartig, Armin Schenk.
Paper describing cito
Citations
cito has been cited 9 times on Google Scholar as of 29 October 2026. These are the citations of the paper describing it. The chart shows how they are distributed over the years, as of 21 October 2026; the 2026 figure covers the year up to that date.
Citations per year
Widely cited work citing cito
The most-cited articles that cite cito or its paper, ranked by their own citation count in OpenAlex as of 29 October 2026; the group's own papers are left out.
- Artificial Intelligence for Climate Change Biology: From Data Collection to Predictions
- Inferring the extinction risk of marine fish to inform global conservation priorities
- Prediction of the volume of shallow landslides due to rainfall using data-driven models
- Neural Network in Aircraft Customer Satisfaction Prediction