Page 17 - Special Topic Session (STS) - Volume 4

STS556 Mohd Bakri A. et al.
Table 1 Performance of existing and modified smoother measured by regression
coefficient

Type of Regression
modification Coefficient

Arithmetic 0.9605

Geometric 0.9622

Quadratic 0.9601

Harmonic 0.9615

Contra harmonic 0.9610

Figure 3 Plot signal versus modified compound smoother using geometric mean

The results, supported by the graphical analysis in Figure 3, demonstrates that
the smoother has the capability to successfully recover the signal from noise
of high volatility. Therefore, the main features of the signal were maintained,
resulting in the further analysis such as model estimation, to be less
complicated.

4. Conclusion
This study is mainly to assess the performance of modified 4253HT in
capturing sinusoidal plus linear trend signal with heavy noise added. Noise
with high volatility was added to the signal and the performances were
measured by recruiting regression coefficient. The results show that modified
4253HT using geometric mean performed the best in extracting signal from
heavy noise. For future works, the performance of proposed adjustment to
compound smoother will be assessed with the inclusion of different types of
signals and noise.

6 | I S I W S C 2 0 1 9