A new strategy for spatial predictive mapping of mineral prospectivity: Automated hyperparameter tuning of random forest approach

Daviran, Mehrdad; Maghsoudi, Abbas; Ghezelbash, Reza; Pradhan, Biswajeet

doi:10.1016/j.cageo.2021.104688

Cited by 68 publications

(22 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Evaluasi kinerja klasifikasi dari model Decision Tree, Random Forest dan Deep Forest pada masing-masing dataset ReLink menggunakan nilai AUC (Area under the ROC Curve). AUC dipilih sebagai metode evaluasi karena AUC cocok untuk mengevaluasi nilai kinerja prediksi yang menggunakan dataset dengan permasalahan kelas tidak seimbang (class imbalance) [24].…”

Section: Evaluasiunclassified

Peningkatan Kinerja Prediksi Cacat Software Dengan Hyperparameter Tuning Pada Algoritma Klasifikasi Deep Forest

et al. 2022

View full text Add to dashboard Cite

Prediksi cacat software adalah salah satu studi pada bidang Rekayasa Perangkat Lunak yang telah diteliti oleh banyak peneliti. Tujuan dari studi ini adalah untuk mencari tahu algoritma yang dapat memberikan kinerja prediksi cacat software yang lebih baik. Salah satu penelitian yang telah dilakukan adalah melakukan prediksi cacat software dengan menggunakan algoritma berbasis pohon seperti Decision Tree, Random Forest dan Deep Forest. Deep Forest adalah algoritma klasifikasi berbasis pohon yang baru yang merupakan perbaikan dari algoritma Random Forest. Namun implementasi Deep Forest dalam penelitian terdahulu masih belum memberikan kinerja yang maksimal. Hasil pada penelitian terdahulu menunjukan bahwa kinerja algoritma Deep Forest masih ada yang lebih rendah dibandingkan algoritma berbasis pohon yang lain. Pada penelitian ini berfokus pada peningkatan kinerja algoritma berbasis pohon dengan melakukan normalisasi pada dataset dan hyperparameter tuning pada algoritma klasifikasi dengan menggunakan pencarian grid. Dataset yang digunakan adalah 3 dataset dari ReLink yaitu Apache, Safe, dan Zxing. Setiap model prediksi divalidasi dengan Stratified 10-Fold Cross Validation dan kinerja dievaluasi menggunakan AUC. Dari hasil eksperimen yang didapatkan,hasil prediksi dari pendekatan yang diusulkan lebih baik daripada metode sebelumnya.

show abstract

Section: Evaluasiunclassified

Peningkatan Kinerja Prediksi Cacat Software Dengan Hyperparameter Tuning Pada Algoritma Klasifikasi Deep Forest

et al. 2022

View full text Add to dashboard Cite

show abstract

“…The performance of the model can be further fine-tuned by varying the set of hyperparameters. The performance of the model is highly sensitive to the values of hyper parameters (Daviran et al, 2021), and several costeffective ways are available, for multi-criteria optimisation (Liu et al, 2017). The number of trees in the forest, the maximum number of features considered for splitting a node, the maximum depth of the tree, and the minimum number of samples used to split a node are varied in this study to improve the efficiency of the model.…”

Section: Machine Learning Algorithm and Performance Evaluationmentioning

confidence: 99%

Proposing an easy-to-use tool for estimating landslide dimensions using a data-driven approach

et al. 2022

Self Cite

View full text Add to dashboard Cite

The increase in population and urbanisation of hilly regions have increased the risk due to landslides. This manuscript presents a data-driven approach with a random forest algorithm to estimate the projected area, length, travel distance, and width of landslides, using elevation and slope information. The method is tested for two different study areas (Idukki and Wayanad), using three different combinations of inputs. The input features considered were elevation (E), tangential slope (θ), drop height (H), angle of reach (α) and the profile curvature (c). A total of 144 models were considered and were evaluated using mean-absolute-error (MAE) and root-meansquare-error (RMSE) values. The results indicate that, by using E and θ alone, the RMSE value in estimating the length values for flow-like landslides in Wayanad was reduced from 472.74 m to 204.64 m. Out of the 48 combinations considered, MAE values have increased in seven cases and RMSE values in eight cases only. The pre-trained models are saved and used to develop an easy-to -use tool, which can bypass the complications associated with the existing statistical approaches. The tool can be used by untrained personnel for preliminary hazard assessment.

show abstract

“…Therefore, in our work, a k-fold cross validation method was used. (23) The k-fold cross validation divides the training data into k equal parts, uses k−1 of the k-divided training data for setting the search range of the hyperparameters, and validates the model performance using the remaining training data as validation data. The validation process is repeated k times to determine the hyperparameter with the lowest generalization error as the final model to which the test data is applied.…”

Section: Predictive Evaluation Using Machine Learning Techniquesmentioning

confidence: 99%

Analysis of Fire Risk Factors in Seoul, Korea, by Machine Learning

Seo¹,

Osorio²,

Yoo³

2022

Sensors and Materials

View full text Add to dashboard Cite

Different types of fire accidents in the urban area of Seoul, Korea are continuously occurring, causing risk and damage to property and life. In this study, we analyze various spatial and nonspatial fire risk factors by applying machine learning techniques to predict their level of importance in future events. We use the data on fire accident for three years (2017-2019) published by the Korean Fire Service and the Seoul Metropolitan Government. Regarding the machine learning techniques, we use support vector machine (SVM), random forest (RF), and gradient boosted regression tree (GBRT). As the first phase, a multiple regression analysis is performed to select seven main factors related to fire occurrence. In the second phase, we calculate the mean absolute error (MAE) and root mean squared error (RMSE) using validation and test data for the machine learning techniques, revealing that RF obtains ideal results. In the third phase, we analyze the importance of the seven fire factors using RF, resulting in the ignition condition (produced by electrical, mechanical, and chemical reasons) being the main factor in fire occurrence. This study is expected to be used as an important guideline to define urban fire reduction and management measures in Seoul, the capital of South Korea.

show abstract

A new strategy for spatial predictive mapping of mineral prospectivity: Automated hyperparameter tuning of random forest approach

Cited by 68 publications

References 33 publications

Peningkatan Kinerja Prediksi Cacat Software Dengan Hyperparameter Tuning Pada Algoritma Klasifikasi Deep Forest

Peningkatan Kinerja Prediksi Cacat Software Dengan Hyperparameter Tuning Pada Algoritma Klasifikasi Deep Forest

Proposing an easy-to-use tool for estimating landslide dimensions using a data-driven approach

Analysis of Fire Risk Factors in Seoul, Korea, by Machine Learning

Contact Info

Product

Resources

About