Statistically Inspired Modification Of Partial Least Square Untuk Memprediksi Suhu Dan Kelembaban Dengan Pra-Pemrosesan Principal Component Analysis

Septiana, Lauda (2014) Statistically Inspired Modification Of Partial Least Square Untuk Memprediksi Suhu Dan Kelembaban Dengan Pra-Pemrosesan Principal Component Analysis. Undergraduate thesis, Institut Teknologi Sepuluh Nopember Surabaya.

[img]
Preview
Text
1310100079-Undergraduate_Thesis.pdf - Published Version

Download (1MB) | Preview

Abstract

Informasi tentang prakiraan cuaca yang cepat dan tepat menjadi hal yang penting dalam aktifitas manusia. BMKG masih menggunakan metode subjektif dalam melakukan prakiraan cuaca jangka pendek. Salah satu informasi cuaca jangka pendek yang diumumkan oleh BMKG yaitu suhu maksimum (TMAKS), suhu minimum (TMIN), dan kelembaban ratarata (RH). Sejak tahun 2004, BMKG memanfaatkan data Numerical Weather Prediction (NWP) sebagai upaya meramalkan cuaca secara objektif, namun data NWP masih menghasilkan ramalan yang bias. Oleh karena itu, perlu dilakukan pemrosesan (post processing) menggunakan metode Model Output Statistics (MOS). MOS merupakan model yang menghubungkan an-tara variabel respon (hasil observasi cuaca) dengan variabel prediktor (parameter NWP) berbasis regresi. Observasi cuaca yang digunakan sebagai variabel respon adalah TMAX, TMIN, dan RH, sedangkan parameter NWP yang digunakan sebanyak 18 variabel. Sebelumnya dilakukan reduksi dimensi dalam sembilan grid pengukuran untuk masing-masing variabel NWP menggunakan Principal Component Analysis (PCA). Metode regresi yang digunakan adalah Statistically Inspired Modification of Partial Least Square (SIMPLS). Sebagian besar komponen utama yang terbentuk dari setiap variabel NWP adalah sebanyak satu komponen. Hasil penelitian menyim-pulkan bahwa validasi model SIMPLS dengan kriteria Root Mean Square Error Prediction (RMSEP) menunjukkan bahwa RMSEP untuk TMAKS di empat stasiun berkriteria sedang. Nilai RMSEP untuk TMIN di tiga stasiun berkriteria baik dan nilai RMSEP untuk RH berkriteria baik di dua stasiun pengamatan. Nilai Percentage Improval (%IM) untuk prediksi TMIN berkisar antara 20,20 - 89,75 persen, yang artinya model SIMPLS dapat mengkoreksi bias NWP antara 20,20 - 89,75 persen ================================================================================================== Information about weather forecasts quickly and accurate to be important in human activities. BMKG was using subjective methods in short-term weather forecast. One of the short-term weather information announced by BMKG is temperature maximum (TMAKS), temperature minimum (TMIN), and humidity (RH). Since 2004, BMKG use Numerical Weather Prediction (NWP) to predict the weather objectively. Therefore, it is necessary to processing (post processing) using Model Output Statistics (MOS). MOS is model that connects between response variable (weather observation) and predictor variable (NWP parameters) based regression. Weather observation were used as the response variabel is TMAX, TMIN, and RH, while NWP parameters are used as many as 18 variables. Previously conducted reduction dimension in nine grid for each variable using Principal Component Analysis (PCA). Regression method used is Statistically Inspired Modification of Partial Least Square (SIMPLS). Most of the principal components that formed of each variable NWP there is one principal component. The regression method used is Statistically Inspired Modification of Partial Least Square (SIMPLS). The result of the study concluded that validation model SIMPLS with criteria Root Mean Square Error Prediction (RMSEP) indicates that RMSEP for TMAKS in four stations have precisely criterion. RMSEP values for TMIN in three stations have good criterion and the value of RMSEP for RH have good criterion in two stations. The value of Percentage Improval (%IM) to predict TMIN between 20,20% until 89,75%, which means that model SIMPLS can correct bias NWP between 20,20% until 89,75%

Item Type: Thesis (Undergraduate)
Additional Information: RSSt 519.535 4 Sep s
Uncontrolled Keywords: MOS, NWP, Principal Component Analysis, SIMPLS, Suhu dan Kelembaban
Subjects: Q Science > QA Mathematics > QA275 Theory of errors. Least squares. Including statistical inference
Q Science > QA Mathematics > QA278.5 Principal components analysis. Factor analysis. Correspondence analysis (Statistics)
Divisions: Faculty of Mathematics and Science > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User: EKO BUDI RAHARJO
Date Deposited: 16 Oct 2020 06:06
Last Modified: 16 Oct 2020 06:37
URI: https://repository.its.ac.id/id/eprint/82157

Actions (login required)

View Item View Item