Prabowo, Nursya`bani Hendro (2017) Penentuan Pelebaran Window Time Optimal pada Data Deret Waktu. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 1313100121_Undergraduate_Theses.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
1313100121_Undergraduate_Theses.pdf Restricted to Repository staff only Download (2MB) | Request a copy |
Abstract
Outlier merupakan kejadian yang secara umum terjadi pada data riil atau dunia nyata. Outlier kerap kali mengandung informasi yang cukup penting sehingga dapat mempengaruhi perubahan model atau data secara signifikan. Keunikan dan pentingnya outlier tersebut menyebabkan kejadian atau kasus outlier sebaiknya tidak dibuang atau dihilangkan dari analisis.
Penelitian yang dilakukan merupakan kelanjutan dari latar belakang penelitian sebelumnya untuk menemukan metode baru dalam mendeteksi outlier. Metode yang telah diinisiasi adalah menggunakan window time dalam mendeteksi outlier dengan data bebas outlier yang telah ditemukan panjang optimalnya. Panjang optimal yang telah diteliti kemudian dipadukan dengan perlunya dilakukan penelitian terhadap pelebaran dari metode window time tersebut antar satu observasi ke observasi lainnya. Penggunaan metode window time dengan pelebaran 1 hingga 10 observasi menghasilkan persentase kesalahan deteksi outlier yang lebih baik serta akurasi untuk melakukan prediksi yang cenderung memiliki keunggulan dibandingkan melakukan pelebaran yang panjang mulai dari 50 observasi atau lebih.
=============================================================================================
Outlier is a general case that happened in real world data. Important information often include in outlier and affecting the model change with significantly. Their uniquely and importance made outlier case better to not remove from the analysis. Some research gave recommendation to change the first model into new model that has been inserting by outlier model.
This research is continuing the past work with background to finding new method on detecting the outlier. The initiated method is to use window time in detecting outliers with free outlier data that have found the optimum length. The optimal length that has been studied then combined with the need for researching on the widening of the window time method between observation one to another. Window time method with widening from observation 1 to 10 resulted in better error detection percentages and more accurate prediction also tended to have advantages rather than over long widening ranging from 50 or more observations.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | Additional outlier, ARIMA, Innovational outlier, Level shift, Outlier detection, Temporary change |
| Subjects: | H Social Sciences > HA Statistics Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) |
| Divisions: | Faculty of Mathematics and Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Nursyabani Hendro Prabowo |
| Date Deposited: | 30 Oct 2017 01:56 |
| Last Modified: | 09 Jan 2018 07:22 |
| URI: | http://repository.its.ac.id/id/eprint/47766 |
Actions (login required)
 |
View Item |