Kecerdasan Buatan Dalam Prediksi Harga Properti: Studi Pemanfaatan Random Forest Untuk Efisiensi Pasar Properti Di Kota Malang

Noorihsan, Sandrian Yulian Firmansyah (2025) Kecerdasan Buatan Dalam Prediksi Harga Properti: Studi Pemanfaatan Random Forest Untuk Efisiensi Pasar Properti Di Kota Malang. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6032241037-Master_Thesis.pdf Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Pasar properti di Kota Malang terus berkembang seiring meningkatnya permintaan perumahan, namun keterbatasan transparansi harga masih menghambat pengambilan keputusan yang terinformasi bagi pembeli, penjual, dan pengembang. Studi ini bertujuan membangun model prediksi harga properti berbasis data menggunakan algoritma Random Forest, yang dikenal mampu memodelkan hubungan non-linier yang kompleks secara stabil. Dataset dikumpulkan melalui web scraping dari Rumah123.com dan diproses melalui serangkaian tahap prapemrosesan, menghasilkan 1.573 observasi bersih. Analisis mengintegrasikan karakteristik properti utama, meliputi variabel temporal (bulan, tahun), atribut fisik (luas tanah, luas bangunan, jumlah kamar tidur, jumlah kamar mandi, kapasitas listrik, dan jumlah lantai), karakteristik properti (sertifikat, jenis properti, kondisi properti, kondisi furnitur, dan posisi hook), serta informasi harga. Model akhir dikembangkan menggunakan parameter optimal dan dievaluasi dengan metrik regresi standar, menghasilkan nilai R² sebesar 76,66%, MAE sebesar Rp367.907.294,81, dan MAPE sebesar 25,27%, yang menunjukkan kemampuannya dalam menangkap pola harga properti di pasar yang dinamis. Temuan ini memberikan implikasi manajerial berupa estimasi harga yang lebih objektif dan berbasis data untuk mendukung keputusan penetapan harga, strategi pemasaran, serta penilaian nilai wajar bagi pengembang, agen, dan calon pembeli.
================================================================================================================================
The property market in Malang City continues to grow along with increasing housing demand, but limited price transparency still hinders informed decision-making for buyers, sellers, and developers. This study aims to build a data-driven property price prediction model using the Random Forest algorithm, known for its ability to stably model complex non-linear relationships. The dataset was collected through web scraping from Rumah123.com and processed through a series of preprocessing stages, resulting in 1,573 clean observations. The analysis integrates key property characteristics, including temporal variables (month, year), physical attributes (land area, building area, number of bedrooms, number of bathrooms, electricity capacity, and number of floors), property characteristics (certificate, property type, property condition, furniture condition, and hook position), and price information. The final model was developed using optimal parameters and evaluated with standard regression metrics, resulting in an R² value of 76.66%, an MAE of Rp367,907,294.81, and a MAPE of 25.27%, demonstrating its ability to capture property price patterns in a dynamic market. These findings provide managerial leverage in the form of more objective, data-driven price estimates to support pricing decisions, marketing strategies, and fair value assessments for developers, agents, and prospective buyers.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Prediksi Harga Properti, Random Forest, Mean Decrease Impurity, Web Scraping, Kota Malang, Property price prediction, Random Forest, Mean Decrease Impurity, Web Scraping, Malang City
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT)
Depositing User:	Sandrian Yulian F Noorihsan
Date Deposited:	23 Jan 2026 07:08
Last Modified:	23 Jan 2026 07:08
URI:	http://repository.its.ac.id/id/eprint/130228

Actions (login required)

View Item