eprintid: 8643
rev_number: 7
eprint_status: archive
userid: 46
dir: disk0/00/00/86/43
datestamp: 2024-10-30 06:12:23
lastmod: 2024-10-30 06:13:01
status_changed: 2024-10-30 06:12:23
type: thesis
metadata_visibility: show
creators_name: Azizan, Muhammad
creators_name: Harsani, Prihastuti
creators_name: Herdianto Situmorang, Boldson
creators_NPM: 065119157
contributors_type: http://www.loc.gov/loc.terms/relators/THS
contributors_type: http://www.loc.gov/loc.terms/relators/THS
contributors_type: http://www.loc.gov/loc.terms/relators/THS
contributors_name: Azizan, Muhammad
contributors_name: Harsani, Prihastuti
contributors_name: Herdianto Situmorang, Boldson
corp_creators: Universitas Pakuan
corp_creators: Fakultas Matematika dan Ilmu Pengetahuan Alam
corp_creators: Program Studi Ilmu Komputer
title: RAINFALL PREDICTION MODEL USING 
MACHINE LEARNING ALGORITHM
ispublished: pub
subjects: QK
divisions: sch_ecs
full_text_status: public
abstract: RAINFALL PREDICTION MODEL USING 
MACHINE LEARNING ALGORITHM
Muhammad Azizan1
, Prihastuti Harsani2
, Boldson H.Situmorang3
1,2,3 Department of Computer Science, Faculty of Mathematics and Natural Science, Pakuan
University, Bogor, West Java, 16143, Indonesia
Abstract
Indonesia is a country where most regions experience two seasons: the dry season and the rainy 
season. The significant difference between these seasons lies in the average monthly rainfall. Rainfall is 
a key indicator in measuring weather conditions, as the amount of measured rainfall can indicate the 
intensity of rain in a specific area. Rainfall is measured in millimeters (mm) at monthly intervals. Rainfall 
has a major impact on human life, both in excess and in shortage. Excessive rainfall, such as heavy or 
extreme rain, often leads to flooding, especially in the city of Bogor, which is known as a flood and 
landslide-prone area due to its high rainfall intensity. Therefore, an effective rainfall prediction method 
is needed to help reduce the negative impacts of extreme weather. In this study, a rainfall prediction model 
was developed for the Bogor area using the Multilayer Perceptron algorithm, with 90% of the data for 
training and 10% for testing. The results show that the model can predict rainfall accurately, with an 
RMSE value of 17.8763, indicating a good level of accuracy. RMSE is used as the main indicator in 
evaluating model performance, where the lower the value, the more accurate the prediction. This MLP�based rainfall prediction model has great potential for application in various sectors, including disaster 
mitigation. With high accuracy, this model can support decision-making in areas vulnerable to floods and 
droughts, thus minimizing the negative impacts of weather variability.
Keywords: Prediction; Rainfall; Multilayer Perceptron
1. Introduction
Indonesia is a country where most of its regions experience two seasons in a year: the dry season and 
the rainy season. The main difference between these two seasons is the average monthly rainfall. Rainfall 
refers to the height of water collected in a rain gauge placed on a flat surface that does not absorb, drain, or 
evaporate [1]. Rainfall plays a significant role in human life on this planet. Both high and low rainfall levels 
greatly impact the climate conditions on Earth’s surface. Excessive rainfall can cause various problems, 
such as floods, crop failures, and other events. Heavy rain with high intensity, commonly referred to as 
extreme rain, can trigger floods. Bogor, one of the cities in Indonesia, often falls victim to flooding due to 
high rainfall, and this sometimes leads to landslides and building damage, causing significant losses to the 
local population. Therefore, a specific method is required to obtain optimal rainfall predictions [2]. 
Artificial Neural Networks, specifically the Multilayer Perceptron (ANN-MLP) model, can be used to
predict weather changes more accurately, enabling the application of this forecasting system in daily life 
[3]. The variables used in the MLP method include rainfall, average temperature, average humidity, and 
sunshine duration. The advantage of using the MLP method lies in its ability to determine better weight 
values compared to other methods [4]. In this study, the selection of the Multilayer Perceptron method is 
expected to improve rainfall prediction accuracy by modeling complex relationships between variables in 
the context of weather data. The goal of this research is to predict rainfall in Bogor in 2023 using data 
downloaded from the BMKG website. The information generated from this research is expected to serve as 
a basis for the Meteorology, Climatology, and Geophysics Agency (BMKG) of Bogor in anticipating the 
future impacts of rainfall.
date: 2024-07-29
date_type: published
pages: 56
institution: Universitas Pakuan
department: Fakultas Matematika dan Ilmu Pengetahuan Alam
thesis_type: Skripsi
thesis_name: Sarjana
citation:   Azizan, Muhammad and Harsani, Prihastuti and Herdianto Situmorang, Boldson  (2024) RAINFALL PREDICTION MODEL USING MACHINE LEARNING ALGORITHM.  Skripsi thesis, Universitas Pakuan.   
document_url: http://eprints.unpak.ac.id/8643/1/Skripsi_065119157_M.Azizan.pdf