| Title |
Prediction using functional regression trees / |
| Translation of Title |
Prognozavimas naudojant funkcinės regresijos medžius. |
| Authors |
Kavaliauskaitė, Miglė |
| Full Text |
|
| Pages |
59 |
| Keywords [eng] |
Regression Energy Tree, functional data, resampling techniques, bagging, repeated training, k-fold cross validation. |
| Abstract [eng] |
Nowadays, the main goal of many models for data analysis is to predict the results as accurately as possible. Machine learning methods help to achieve better results. However, only a few methods are adapted for using functional data. One of them is the Regression Energy Tree. This regression tree using framework of decision tree where the testing procedure is implemented using energy statistics such as distance correlation. This paper deals with scalar-on-function regression where the regressors are curves and the response variable is scalar. In this work, the Regression Energy Tree is complemented by different resampling techniques such as bagging, repeated training, and k-fold cross validation. Tests have shown that using any of the resampling techniques presented in the paper, the model predicts more accurately than without using resampling techniques. |
| Dissertation Institution |
Vilniaus universitetas. |
| Type |
Master thesis |
| Language |
English |
| Publication date |
2022 |
