Construction of symmetry triangular fuzzy number procedure (STFNP) using statistical information for autoregressive forecasting
Single-point data are used for data collection. However, data collected by various data collection methods are often exposed to uncertainties that may affect the information presented by the quantitative results. This also causes the forecast model developed to be less precise because of the uncerta...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English English |
| Published: |
2020
|
| Subjects: | |
| Online Access: | http://eprints.uthm.edu.my/403/1/24p%20MUHAMMAD%20SHUKRI%20CHE%20LAH.pdf http://eprints.uthm.edu.my/403/2/MUHAMMAD%20SHUKRI%20CHE%20LAH%20COPYRIGHT%20DECLARATION.pdf http://eprints.uthm.edu.my/403/3/MUHAMMAD%20SHUKRI%20CHE%20LAH%20WATERMARK.pdf http://eprints.uthm.edu.my/403/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Single-point data are used for data collection. However, data collected by various data collection methods are often exposed to uncertainties that may affect the information presented by the quantitative results. This also causes the forecast model developed to be less precise because of the uncertainties contained in the input data. It is essential to describe the uncertainty in data to obtain a realistic result from data analysis. However, most studies focus on model uncertainty regardless of data uncertainty. The data processing carried out may not always take care of uncertainty. When uncertainties in the raw data are not sufficiently handled, this creates more errors that are included in the predicted model. Standard procedures are also very limited to be followed in order to transform a single-point value into Triangular Fuzzy Number (TFN), which addresses the uncertainty. Thus, the data preparation procedure of Symmetry Triangular Fuzzy Number (STFN) is presented in this study to build an improved autoregressive model for time series forecasting. This study presents the proposed Symmetry Triangular Fuzzy Number Procedure (STFNP) using percentage error method and standard deviation method for first-order autoregressive forecasting. Percentage error rate method involves three different percentage rates, while the second method uses the standard deviation of the data. Simulations and verification procedures are presented and are accompanied with numerical examples using actual datasets of Air Pollutant Index and stock markets of selected ASEAN countries. This study reveals that the percentage error and standard deviation methods, which were used to construct the TFN, can achieve the same or better accuracy as compared to a single-point procedure. The results of the simulations and experiments show that the standard deviation method produces better results compared to the other proposed approaches and the conventional approach. Besides, the systematic procedure to construct the TFN does not deviate from single-point procedures. Importantly, uncertain data being treated avoids more uncertainties that would have been brought to the outcome of the forecast model and consequently improves prediction accuracy. |
|---|