13.05.2018 Views

merged

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Outlier Detection by Using Robust Regression Parameter Estimation<br />

Nurul Najwa Binti Abdullah<br />

Supervisor: Mrs Nor Azlida Binti Aleng<br />

Bachelor of Science (Financial Mathematics)<br />

School of Information and Applied Mathematics<br />

The presence of outliers in a data set can result in significant error rates. Outlier that<br />

exists in some data sets can be important data for the data set or may be non-critical<br />

data in a data set. The study was conducted using 500 blood pressure data sample The<br />

purpose of this study was to investigate the presence of outlier in blood pressure data<br />

sample, making comparisons on some of the best methods of detecting outlier and to<br />

test the robustness of sampled data when contaminated by 10%, 30% and 50%. The<br />

method used in this study was a robust regression method using the Least Trimmed<br />

Square (LTS) estimator and the MM estimator. The outlier was detected by using the SAS<br />

9.4. The findings show that outlier detection using robust regression method which is LTS<br />

estimators is more effective than MM estimators.<br />

879 | UMT UNDERGRADUATE RESEARCH DAY 2018

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!