Search
Abstract
Nowadays, Machine-Learning (ML) methods can help engineers solve issues using developed and well-trained computer algorithms. It is possible to predict the seismic behavior of steel Moment-Resisting Frames (MRFs) with higher numerical accuracy. The present research aims to propose a Feature Selection Method (FSM) that can be employed by ML methods to compensate for the "not given feature/features", which are not possible to be prepared for several reasons (e.g., in real conditions). It is noteworthy that the FSM has been developed in Python programming language to do all steps automatically with a pre-defined number of features that need to be predicted and added to the test dataset. Since each of the features has the corresponding partial dependence in the prediction model, it is necessary to keep the features used for the training dataset, and then, the same number of features for the testing dataset. To justify the generality of the proposed FSM, the 2-, and 6-Story MRFs were selected, and the training dataset used for maximum Interstory Drift (ID) prediction has been considered. The results show that the FSM has the ability to generate not-given features and achieve acceptable numerical prediction. For example, the proposed FSM can predict the five not-given features of the test dataset of the 2-, and 6-Story MRFs, and the R2 values of 94.7% and 93.4% have been determined based on the predicted test dataset.
Authors (6)
Cite as
Full text
full text is not available in portal
Keywords
Details
- Category:
- Conference activity
- Type:
- publikacja w wydawnictwie zbiorowym recenzowanym (także w materiałach konferencyjnych)
- Language:
- English
- Publication year:
- 2024
- Bibliographic description:
- Asgarkhani N., Kazemi F., Manguri A., Jakubczyk-Gałczyńska A., Lasowicz N., Jankowski R.: IMPROVING COMPUTATIONAL ABILITY OF MACHINE-LEARNING ALGORITHMS WITH FEATURE SELECTION METHOD// / : , 2024,
- Sources of funding:
-
- Free publication
- Verified by:
- Gdańsk University of Technology
seen 0 times