BlueML: machine learning for metocean models

[🚧 Pre-release version - expect breaking changes 🛠️]

The library for finetuning your metocean model results to real world observations.

Getting Started

pip install git+https://github.com/DHI/blue_ml-preview.git

Example Workflow

• Notebook: Workflow quick start

Start by defining a Timeseries object with your data.

import blue_ml
import pandas as pd
timeseries = blue_ml.Timeseries(
   features = pd.read_csv("coarse_model_data.csv")
   targets = pd.read_csv("observed_hm0.csv")
)

Setup a ModelFrame with your choice of data transformations and machine learning architecture.

model_frame = (
    blue_ml.ModelFrame("Demo ML Framework")
    .add_model(blue_ml.machinelearning.architectures.BlueDense())
    .add_scaler(blue_ml.transforms.StandardScaler())
)

Fit and evaluate the model

model_skill = model_frame.fit_evaluate(timeseries)
model_skill.plot_scatter()

Publications

License

This project is licensed under the MIT License - see the LICENSE file for details.

Support

For questions, issues, or contributions, please visit our GitHub repository or open an issue.