Chen et al., 2023. Assessing the conservation status of Chinese freshwater fish using deep learning
Contact: chen_jinnan@126.com
Update: 1 June, 2023

This repository contains the code and database used in the paper 'Assessing the conservation status of Chinese freshwater fish using deep learning'. 

## IUCNN folder:

The IUCNN folder is the R language project folder and contains the optimal model code used for assessing the conservation status of Chinese freshwater fish and the associated input data and output file. 

"iucn_five_nnoptimal.R" : the optimal models for the 5-class evaluation
"iucn_two_nnoptimal.R" : the optimal models for the 2-class evaluation

The "input" and "feature_extraction" folders represent input data, including occurrence data of Chinese freshwater fish, IUCN RedList assessment, biological traits, phylogenetic relationship data, and environmental factor data.

## testing_result folder:

The testing_result folder is the model test results and consists of two types of files, i.e., ".xlsx" and "folder". The file naming principle is "labels' type_model type_input features". For example, "2labels_nn-class_GEO" means 2-class model, model algorithm is nn-class, and input features are GEO (geographic information).


## Abbreviations:
BT, biological traits; 
ECO, proxies for eco-environment; 
GEO, geographic information; 
GEOB, geographic sampling bias; 
HFI, human footprint index; 
PHY, phylogenetic status; 
BIOM, presence of the species across different biomes; 
cnn-class, convolutional neural network class model; 
nn-class, neural network class model; 
nn-reg, neural network regression model; 
bnn-class Bayesian neural network class model.