Hybrid modelling using decision tree and ordered regression: an application to health sciences research
With the increasing complexity of healthcare data, there is a need for more advanced and integrative predictive modelling techniques. This thesis presents a novel hybrid methodology integrating Decision Trees and Ordinal Regression using the R-syntax. The study objectives include the development of...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2024
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/60708/1/HAZIK%20BIN%20SHAHZAD-FINAL%20THESIS%20P-SGD001120%28R%29-E.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | With the increasing complexity of healthcare data, there is a need for more advanced and integrative predictive modelling techniques. This thesis presents a novel hybrid methodology integrating Decision Trees and Ordinal Regression using the R-syntax. The study objectives include the development of the hybrid method, measuring its efficacy and efficiency, validating its performance through predictive classification analysis, and optimising parameter estimates for optimised statistical inferences. The hybrid methodology uses decision trees, facilitated by visualisation tools, to identify influential factors that shape the model’s predictions. The bootstrap resampling method boosts the data set’s resilience and facilitates the development of an ordinal regression model. The introduction of the hybrid accuracy index enhances interpretability. The hybrid methodology is employed in two health sciences scenarios. In Case I, it predicts the frequency of toothbrushing among students, and in Case II, it predicts diabetic status using oral health indicators. This study introduces a hybrid method that generates numerical results along with graphical visualisation, enhancing the accuracy and efficiency of the parameter estimates. The findings of this study contribute to the development of an innovative approach to transforming predictive modelling in healthcare, contributing to future research methodologies for more precise decision-making. |
|---|