imbaterry11/nyus.2

NYUS.2 is an auto machine learning-empowered prediction model for grapevine freezing tolerance

NYUS.2

NYUS.2 is an automated machine learning-empowered prediction model for grapevine freezing tolerance. This repo includes all the source code for feature extraction, model training and model deployment along with the original training data as parts of the open-source study.

Folders and files description

Raw data processing and feature extraction

Feature_extraction.R is an R script to extract features from daily temperature for the training and prediction of NYUS.2
Cultivars.Rdata contains all the names for the Boolean-type cultivar columns
daily_temperature_data_example.csv is an example file to be processed by Feature_extraction.R to generate features
daily_temperature_data_example_feature_extracted.csv is a resulting file that contains necessary features

Training and feature importance quantification

Autogluon_model_training_feature_importance.ipynb is a notebook for the training of NYUS.2 and feature importance quantification
All_training_data_9_sites.csv is the entire LT50 dataset contributed by nine research facilities for NYUS.2 model training and testing

Using model

NYUS.2_using_the_model.ipynb is a notebook for the prediction using NYUS.2
daily_temperature_data_example_feature_extracted.csv is an example file that contains necessary features for prediction
LT50_pred.csv is the model prediction

Additional information:

Ready-to-use model can be download at here.
The current model was deployed at the Cornell grape freezing tolerance prediction app:

Upon the use of the tools provided in this repo, please cite:
Wang, Hongrui, Gaurav D Moghe, Al P Kovaleski, Markus Keller, Timothy E Martinson, A Harrison Wright, Jeffrey L Franklin, et al. 2023. “NYUS.2: An Automated Machine Learning Prediction Model for the Large-Scale Real-Time Simulation of Grapevine Freezing Tolerance in North America.” Horticulture Research, December, uhad286. https://doi.org/10.1093/hr/uhad286.