Model Output Statistics dengan CART dan Random Forests untuk Prakiraan Curah Hujan Harian

., Nurhayati (2017) Model Output Statistics dengan CART dan Random Forests untuk Prakiraan Curah Hujan Harian. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img]
Preview
Text
1315105023-Undergraduate_Theses.pdf - Published Version

Download (1MB) | Preview

Abstract

Cuaca memiliki pengaruh yang sangat penting terhadap kehidupan manusia. Upaya meminimalkan dampak bencana akibat cuaca/iklim maka dibutuhkan informasi prakiraan cuaca/iklim. Khususnya pada bidang transportasi membutuhkan informasi cuaca pendek yang sangat cepat dan tepat. Badan Meteorologi, Klimatologi, dan Geofisika (BMKG) sebagai lembaga resmi pemerintah yang bertugas memberikan informasi cuaca. Salah satu informasi cuaca jangka pendek yang cukup penting adalah curah hujan. BMKG telah berupaya melakukan pemodelan cuaca jangka pendek dengan memanfaatkan data komponen cuaca Numerical Weather Prediction (NWP). Salah satu upaya untuk mengoptimalkan pemanfaatan output model NWP adalah dengan Model Output Statistics (MOS), karena prakiraan NWP masih bias. Metode yang akan digunakan untuk pendekatan MOS adalah Classification and Regression Tree (CART) dan random forests. Tujuan penelitian ini adalah mengetahui hasil ketepatan klasifikasi curah hujan harian melalui metode tersebut. Hasil penelitian menunjukkan bahwa ketepatan klasifikasi untuk metode CART pada Stasiun Meteorologi Juanda, Stasiun Meteorologi Ngurai Rai dan pada Stasiun Meteorologi Soekarno-Hatta (Soetta) yaitu: 46.67%, 73.33% dan 66.67%. Sedangkan untuk metode random forests yaitu: 53.33%, 53.33% dan 70.00%.Metode random forests mampu meningkatkan akurasi hasil prakiraan curah hujan harian. ============================================================================== Weather has a very important influence on human life. The efforts to minimize the impact of disasters which caused by weather / climate therefor weather / climate forecasting information is required. Especially in the transportation field which requires a fast and precise of short weather information. Indonesian agency for Meteorological, Climatological and Geophysics (BMKG) as the official government agency is in charge of providing weather information. One of the important short-term weather information is precipitation. BMKG has attempted short-term weather modeling using Numerical Weather Prediction (NWP) weather component data. One of efforts to optimize the utilization of NWP model output is by Model Output Statistics (MOS), as NWP forecasts are still biased. The methods to be used for the MOS approach are Classification and Regression Tree (CART) and random forests. The purpose of this research is to know the accuracy result of daily precipitation classification through the method. The results showed that the accuracy of classification for CART method at Juanda Meteorological Station, Ngurai Rai Meteorology Station and Soekarno-Hatta Meteorological Station (Soetta) are 46.67%, 73.33% and 66.67%. Whereas for random forests method are: 53.33%, 53.33% and 70.00%. Random forests method can improve the accuracy of daily precipitation forecast results.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: CART, Curah Hujan Harian, MOS, NWP, Random Forests, precipitation daily
Subjects: H Social Sciences > HA Statistics
Q Science > QA Mathematics > QA278.5 Principal components analysis. Factor analysis.
Divisions: Faculty of Mathematics and Science > Statistics > (S1) Undergraduate Theses
Depositing User: Nurhayati . .
Date Deposited: 05 Oct 2017 05:03
Last Modified: 06 Mar 2019 07:31
URI: http://repository.its.ac.id/id/eprint/44884

Actions (login required)

View Item View Item