🌱 Applying-XAI-approaches-to-ecology-Shiny

Overview 📜

This Shiny app accompanies the study titled "Applying an interpretable machine learning approach to assess intraspecific trait variation under landscape-scale population differentiation," authored by Sambadi Majumder and Dr. Chase Mason. The study, accessible through this DOI link, investigates the functional trait data of Helianthus annuus genotypes from the HeliantHome database.

Data Source 🌐

The functional trait data is from the HeliantHome database, publicized in Bercovich et al., 2022 (DOI link). Genotypes were selected based on their occurrence within the Level I ecoregions of the Great Plains and North American Deserts, correlated with ecoregion shapefiles from the U.S. Environmental Protection Agency.

Built With 🛠️

• Programming Language: R
• Key Packages:
  • shiny for app functionality.
  • leaflet for interactive mapping 🗺️.
  • ggplot2 and plotly for creating visualizations 📊.
  • dplyr for data manipulation 🔨.
  • sf for handling spatial data 🌍.

App Structure 📚

Study Region 📍

Displays a map of Helianthus annuus populations' distribution within the Great Plains and North American Deserts ecoregions.

Divergent Traits 🔍

Reveals traits exhibiting divergence between the Desert and Plains populations, highlighting those most predictive of ecoregion.

Impacts on Divergence 🔬

Shows accumulated local effects plots articulating the impact of each divergent trait on ecoregion classification.

Research and Application 🧬

Demonstrates an interpretable machine learning approach for identifying ecoregion-predictive traits, essential for ecological strategy research in Helianthus annuus.

How to Use 🤔

Navigate between study components through an intuitive interface, engaging with visualizations for a deeper understanding of the findings.

Explore the interactive visualizations on the Shiny app here.