Ensemble Machine Learning of Random Forest, AdaBoost and XGBoost for Vertical Total Electron Content Forecasting

Natraš, Randa; Soja, Benedikt; Schmidt, Michaël

doi:10.3390/rs14153547

Cited by 55 publications

(48 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Adapun beberapa riset terbaru yang menggabungkan penggunaan metode AdaBoost dan random forest dapat ditemukan pada [13][14][15]. Hasil riset dengan kedua metode tersebut banyak dilakukan dalam berbagai kasus sehingga menarik untuk diterapkan pada kasus lain terutama pada kasus program JKN-KIS dengan memanfaatkan big data yang dimiliki oleh BPJS Kesehatan.…”

Section: Pendahuluanunclassified

Metode AdaBoost dan Random Forest untuk Prediksi Peserta JKN-KIS yang Menunggak

2023

View full text Add to dashboard Cite

The contribution of participants, employers, and/or the government is one of the most important things in the National Health Insurance Program-Healthy Indonesia Card (JKN-KIS) implementation. All Indonesian residents were required to participate in the JKN-KIS program which is divided into four types of participation, one of which is Non-Wage Recipient Participants (PBPU) whose contributions are paid independently. However, based on December 2021 data, 60% of PBPU participants were late in paying monthly until they were in arrears. Arrears in payment of contributions cause several problems, including payment of claims to deficits. This research utilized big data owned by the Healthcare and Social Security Agency (BPJS Kesehatan) and machine learning based on ensemble trees, namely AdaBoost and random forest to get the predictions of participants in arrears. The results showed that machine learning based on an ensemble tree was able to predict PBPU participants in arrears with high accuracy, as evidenced by the AUC values in both models above 80%. The random forest model has an F1-score and the AUC value is better than the AdaBoost, namely the F1-score of 85,43% and the AUC value of 87,20% in predicting JKN-KIS participants who are in arrears in payment of contributions.

show abstract

Section: Pendahuluanunclassified

Metode AdaBoost dan Random Forest untuk Prediksi Peserta JKN-KIS yang Menunggak

2023

View full text Add to dashboard Cite

show abstract

“…, whereas the columns represent the input features 𝐴𝐴 x𝐱𝑝𝑝 with p = {0, 1, 2, …, P − 1} (Natras et al, 2022a).…”

Section: Datamentioning

confidence: 99%

“…To model the solar-terrestrial processes and the impact of space weather on VTEC, data on solar and geomagnetic activity were downloaded from the OMNIWeb NASA Service and added as input features. The data set is prepared with a 1-hr resolution, denoted D1, corresponding to Table 1 of Natras et al (2022a). It consists of the following input features of x i at timestamp i: The input data for artificial neural networks were standardized to obtain data with a mean of zero and a standard deviation of one.…”

Section: Datamentioning

confidence: 99%

“…Other studies on ML-based VTEC modeling and forecasting, such as Y. Han et al (2022), Kaselimi et al (2022), L. , Lee et al (2020), to name a few, have not quantified the uncertainties or provided confidence intervals (CI) of VTEC output, leading to a lack of information on how certain and reliable their ML-based VTEC results are; Natras et al (2022a) provides an overview of these studies and their results.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Uncertainty Quantification for Machine Learning‐Based Ionosphere and Space Weather Forecasting: Ensemble, Bayesian Neural Network, and Quantile Gradient Boosting

Natras,

Soja,

Schmidt

2023

Space Weather

Self Cite

View full text Add to dashboard Cite

Machine learning (ML) has been increasingly applied to space weather and ionosphere problems in recent years, with the goal of improving modeling and forecasting capabilities through a data‐driven modeling approach of nonlinear relationships. However, little work has been done to quantify the uncertainty of the results, lacking an indication of how confident and reliable the results of an ML system are. In this paper, we implement and analyze several uncertainty quantification approaches for an ML‐based model to forecast Vertical Total Electron Content (VTEC) 1‐day ahead and corresponding uncertainties with 95% confidence intervals (CI): (a) Super‐Ensemble of ML‐based VTEC models (SE), (b) Gradient Tree Boosting with quantile loss function (Quantile Gradient Boosting, QGB), (c) Bayesian neural network (BNN), and (d) BNN including data uncertainty (BNN + D). Techniques that consider only model parameter uncertainties (a and c) predict narrow CI and over‐optimistic results, whereas accounting for both model parameter and data uncertainties with the BNN + D approach leads to a wider CI and the most realistic uncertainties quantification of VTEC forecast. However, the BNN + D approach suffers from a high computational burden, while the QGB approach is the most computationally efficient solution with slightly less realistic uncertainties. The QGB CI are determined to a large extent from space weather indices, as revealed by the feature analysis. They exhibit variations related to daytime/nightime, solar irradiance, geomagnetic activity, and post‐sunset low‐latitude ionosphere enhancement.

show abstract

“…With the rapid development of artificial intelligence (AI), intelligent methods based on machine learning have been introduced into the modeling of TEC and significantly improved the model's accuracy (Han et al., 2022). Typical machine learning algorithms such as support vector regression (SVR) (Zhukov et al., 2018), nearest neighbor algorithm (Monte‐Moreno et al., 2022), and decision tree method (Natras et al., 2022) all show good performance in ionospheric TEC prediction, effectively improving the prediction accuracy of existing TEC maps, simplifying the prediction process, and reducing the amount of calculation. Artificial neural networks (ANN) can model the system with less information.…”

Section: Introductionmentioning

confidence: 99%

A High Accuracy Spatial Reconstruction Method Based on Surface Theory for Regional Ionospheric TEC Prediction

Wang,

Liu,

Shi

2023

Space Weather

View full text Add to dashboard Cite

In order to achieve more accurate spatial reconstruction of ionospheric total electron content (TEC) and promote improved satellite positioning and ranging applications, a high accuracy spatial reconstruction (HASR) method for TEC is proposed based on the surface theory. The core theory of this method is as follows: (a) Any surface can be uniquely determined by its first and second fundamental quantities; (b) By direct difference approximation, differential equations are transformed into algebraic equations to solve Gauss equations faster. At the same time, taking parts of Europe as an example, the proposed HASR method is used to determine the correlation coefficients and the number of iterations of the model by using the relative root mean square error (RRMSE) as the evaluation criterion. The statistical results show that the TEC predicted by the HASR method is highly consistent with the actual observed values of ionospheric observation stations, and the prediction RRMSE is 9.75%. Compared with the Kriging interpolation with scale factor, the prediction accuracy of the HASR method is improved by 8.5%. We hope this method can provide ideas for the spatial reconstruction of other ionospheric parameters and further promote the realization of complete and accurate space weather forecast.

show abstract

Ensemble Machine Learning of Random Forest, AdaBoost and XGBoost for Vertical Total Electron Content Forecasting

Cited by 55 publications

References 53 publications

Metode AdaBoost dan Random Forest untuk Prediksi Peserta JKN-KIS yang Menunggak

Metode AdaBoost dan Random Forest untuk Prediksi Peserta JKN-KIS yang Menunggak

Uncertainty Quantification for Machine Learning‐Based Ionosphere and Space Weather Forecasting: Ensemble, Bayesian Neural Network, and Quantile Gradient Boosting

A High Accuracy Spatial Reconstruction Method Based on Surface Theory for Regional Ionospheric TEC Prediction

Contact Info

Product

Resources

About