Spatial rain probabilistic prediction performance using costsensitive learning algorithm
The use of machine learning in weather prediction is growing rapidly as an alternative to conventional numerical weather prediction. However, predictions using machine learning such as Long Short Term Memory (LSTM) based on neural networks have weaknesses in predicting extreme events with a high rat...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
EDP Sciences
2023-01-01
|
| Series: | E3S Web of Conferences |
| Online Access: | https://www.e3s-conferences.org/articles/e3sconf/pdf/2023/101/e3sconf_icdmm2023_19001.pdf |
| _version_ | 1797345014450225152 |
|---|---|
| author | Saputra Agung Hari Satya I. Made Agus Sari Fitria Puspita Mulya Aditya |
| author_facet | Saputra Agung Hari Satya I. Made Agus Sari Fitria Puspita Mulya Aditya |
| author_sort | Saputra Agung Hari |
| collection | DOAJ |
| description | The use of machine learning in weather prediction is growing rapidly as an alternative to conventional numerical weather prediction. However, predictions using machine learning such as Long Short Term Memory (LSTM) based on neural networks have weaknesses in predicting extreme events with a high ratio of unbalanced data. This research examines the performance of using focal loss in LSTM to obtain a machine-learning model that is cost-sensitive. The model used the Global Forecasting System Data and the Global Satellite Measurement of Precipitation for the years 2017-2020. Testing the hyperparameter configuration was carried out using the hyperband method on the number of nodes and the number of iterations with 3 scenarios (2, 3, and 4 classes). The results showed an increased performance against noncost sensitive LSTM with an average increase of 25% accuracy and 11% F1-score on 2 classes scenario, 15% accuracy increase and 21% F1-score for scenario 3 classes, as well as an increase in accuracy of 15% and F1-score 26% for scenario 4 class. It also provides the idea of how cost-sensitive properties can help machine learning models detect classes with extreme ratios, based on an increase in average performance as the number of classification scenarios increases. |
| first_indexed | 2024-03-08T11:11:18Z |
| format | Article |
| id | doaj.art-4dfc0a6099c644c885ed13447086963a |
| institution | Directory Open Access Journal |
| issn | 2267-1242 |
| language | English |
| last_indexed | 2024-03-08T11:11:18Z |
| publishDate | 2023-01-01 |
| publisher | EDP Sciences |
| record_format | Article |
| series | E3S Web of Conferences |
| spelling | doaj.art-4dfc0a6099c644c885ed13447086963a2024-01-26T10:41:46ZengEDP SciencesE3S Web of Conferences2267-12422023-01-014641900110.1051/e3sconf/202346419001e3sconf_icdmm2023_19001Spatial rain probabilistic prediction performance using costsensitive learning algorithmSaputra Agung Hari0Satya I. Made Agus1Sari Fitria Puspita2Mulya Aditya3Meteorology Department, The State College of Meteorology Climatology and GeophysicsKotabaru Meteorological Station, Meteorology, Climatology and Geophysics AgencyMeteorology Department, The State College of Meteorology Climatology and GeophysicsMeteorology Department, The State College of Meteorology Climatology and GeophysicsThe use of machine learning in weather prediction is growing rapidly as an alternative to conventional numerical weather prediction. However, predictions using machine learning such as Long Short Term Memory (LSTM) based on neural networks have weaknesses in predicting extreme events with a high ratio of unbalanced data. This research examines the performance of using focal loss in LSTM to obtain a machine-learning model that is cost-sensitive. The model used the Global Forecasting System Data and the Global Satellite Measurement of Precipitation for the years 2017-2020. Testing the hyperparameter configuration was carried out using the hyperband method on the number of nodes and the number of iterations with 3 scenarios (2, 3, and 4 classes). The results showed an increased performance against noncost sensitive LSTM with an average increase of 25% accuracy and 11% F1-score on 2 classes scenario, 15% accuracy increase and 21% F1-score for scenario 3 classes, as well as an increase in accuracy of 15% and F1-score 26% for scenario 4 class. It also provides the idea of how cost-sensitive properties can help machine learning models detect classes with extreme ratios, based on an increase in average performance as the number of classification scenarios increases.https://www.e3s-conferences.org/articles/e3sconf/pdf/2023/101/e3sconf_icdmm2023_19001.pdf |
| spellingShingle | Saputra Agung Hari Satya I. Made Agus Sari Fitria Puspita Mulya Aditya Spatial rain probabilistic prediction performance using costsensitive learning algorithm E3S Web of Conferences |
| title | Spatial rain probabilistic prediction performance using costsensitive learning algorithm |
| title_full | Spatial rain probabilistic prediction performance using costsensitive learning algorithm |
| title_fullStr | Spatial rain probabilistic prediction performance using costsensitive learning algorithm |
| title_full_unstemmed | Spatial rain probabilistic prediction performance using costsensitive learning algorithm |
| title_short | Spatial rain probabilistic prediction performance using costsensitive learning algorithm |
| title_sort | spatial rain probabilistic prediction performance using costsensitive learning algorithm |
| url | https://www.e3s-conferences.org/articles/e3sconf/pdf/2023/101/e3sconf_icdmm2023_19001.pdf |
| work_keys_str_mv | AT saputraagunghari spatialrainprobabilisticpredictionperformanceusingcostsensitivelearningalgorithm AT satyaimadeagus spatialrainprobabilisticpredictionperformanceusingcostsensitivelearningalgorithm AT sarifitriapuspita spatialrainprobabilisticpredictionperformanceusingcostsensitivelearningalgorithm AT mulyaaditya spatialrainprobabilisticpredictionperformanceusingcostsensitivelearningalgorithm |