search

informatics-lab / precip_rediagnosis

Project to use ML to re-diagnose precipitation fields from ensemble model fields
0 stars 0 forks source link

Fraction pipeline #107

Closed hannahbrown7 closed 1 year ago

hannahbrown7 commented 1 year ago

Development of radar precip intensity band fraction modelling, trained on extended dataset closes #45, closes #64, closes #65, closes #54

Function in fraction_model_dev_reduced_bands_azml.ipynb for calculating u and v wind fields - could be incorporated into data prep. closes #93

Updated train/test sampling to ensure all ensemble members for same time and space are in the same sample closes #82

Updated prd_pipelines.py and prd_cluster_train_demo.py to generalise to both fractional and mean precip rate cases. Therefore merging fraction_model_pipeline and model_pipeline folders. closes #103

Exploration of approaches to reduce class imbalance including resampling of different bands and weighting bands when training model. Further assessment required of best approach.

review-notebook-app[bot] commented 1 year ago

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB