A systematic hybrid machine learning approach for stress prediction
Stress is becoming an increasingly prevalent health issue, seriously affecting people and putting their health and lives at risk. Frustration, nervousness, and anxiety are the symptoms of stress and these symptoms are becoming common (40%) in younger people. It creates a negative impact on human liv...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
PeerJ Inc.
2023-02-01
|
| Series: | PeerJ Computer Science |
| Subjects: | |
| Online Access: | https://peerj.com/articles/cs-1154.pdf |
| _version_ | 1811166584138366976 |
|---|---|
| author | Cheng Ding Yuhao Zhang Ting Ding |
| author_facet | Cheng Ding Yuhao Zhang Ting Ding |
| author_sort | Cheng Ding |
| collection | DOAJ |
| description | Stress is becoming an increasingly prevalent health issue, seriously affecting people and putting their health and lives at risk. Frustration, nervousness, and anxiety are the symptoms of stress and these symptoms are becoming common (40%) in younger people. It creates a negative impact on human lives and damages the performance of each individual. Early prediction of stress and the level of stress can help to reduce its impact and different serious health issues related to this mental state. For this, automated systems are required so they can accurately predict stress levels. This study proposed an approach that can detect stress accurately and efficiently using machine learning techniques. We proposed a hybrid model (HB) which is a combination of gradient boosting machine (GBM) and random forest (RF). These models are combined using soft voting criteria in which each model’s prediction probability will be used for the final prediction. The proposed model is significant with 100% accuracy in comparison with the state-of-the-art approaches. To show the significance of the proposed approach we have also done 10-fold cross-validation using the proposed model and the proposed HB model outperforms with 1.00 mean accuracy and +/−0.00 standard deviation. In the end, a statistical T-test we have done to show the significance of the proposed approach in comparison with other approaches. |
| first_indexed | 2024-04-10T15:54:57Z |
| format | Article |
| id | doaj.art-630edef4c6dc4672a5519629918c2c32 |
| institution | Directory Open Access Journal |
| issn | 2376-5992 |
| language | English |
| last_indexed | 2024-04-10T15:54:57Z |
| publishDate | 2023-02-01 |
| publisher | PeerJ Inc. |
| record_format | Article |
| series | PeerJ Computer Science |
| spelling | doaj.art-630edef4c6dc4672a5519629918c2c322023-02-10T15:05:04ZengPeerJ Inc.PeerJ Computer Science2376-59922023-02-019e115410.7717/peerj-cs.1154A systematic hybrid machine learning approach for stress predictionCheng Ding0Yuhao Zhang1Ting Ding2Emory University, Atlanta, GA, United StatesUniversity of Nottingham, Nottingham, United KingdomEast China University of Technology, NAN Chang, ChinaStress is becoming an increasingly prevalent health issue, seriously affecting people and putting their health and lives at risk. Frustration, nervousness, and anxiety are the symptoms of stress and these symptoms are becoming common (40%) in younger people. It creates a negative impact on human lives and damages the performance of each individual. Early prediction of stress and the level of stress can help to reduce its impact and different serious health issues related to this mental state. For this, automated systems are required so they can accurately predict stress levels. This study proposed an approach that can detect stress accurately and efficiently using machine learning techniques. We proposed a hybrid model (HB) which is a combination of gradient boosting machine (GBM) and random forest (RF). These models are combined using soft voting criteria in which each model’s prediction probability will be used for the final prediction. The proposed model is significant with 100% accuracy in comparison with the state-of-the-art approaches. To show the significance of the proposed approach we have also done 10-fold cross-validation using the proposed model and the proposed HB model outperforms with 1.00 mean accuracy and +/−0.00 standard deviation. In the end, a statistical T-test we have done to show the significance of the proposed approach in comparison with other approaches.https://peerj.com/articles/cs-1154.pdfMachine learningStress detectionHybrid appraoch |
| spellingShingle | Cheng Ding Yuhao Zhang Ting Ding A systematic hybrid machine learning approach for stress prediction PeerJ Computer Science Machine learning Stress detection Hybrid appraoch |
| title | A systematic hybrid machine learning approach for stress prediction |
| title_full | A systematic hybrid machine learning approach for stress prediction |
| title_fullStr | A systematic hybrid machine learning approach for stress prediction |
| title_full_unstemmed | A systematic hybrid machine learning approach for stress prediction |
| title_short | A systematic hybrid machine learning approach for stress prediction |
| title_sort | systematic hybrid machine learning approach for stress prediction |
| topic | Machine learning Stress detection Hybrid appraoch |
| url | https://peerj.com/articles/cs-1154.pdf |
| work_keys_str_mv | AT chengding asystematichybridmachinelearningapproachforstressprediction AT yuhaozhang asystematichybridmachinelearningapproachforstressprediction AT tingding asystematichybridmachinelearningapproachforstressprediction AT chengding systematichybridmachinelearningapproachforstressprediction AT yuhaozhang systematichybridmachinelearningapproachforstressprediction AT tingding systematichybridmachinelearningapproachforstressprediction |