Apartment Listings - KASMOM Project

Team Members

• Dafi Nafidz Radhiyya
• Ivan Rabbani C.
• Fariz Eda
• Taqiya Zayin Hanafie

Table of Contents

1. Background & Objectives
2. Dataset Description
3. Tasks
   • Exploratory Data Analysis (EDA)
   • Classification
   • Regression
   • Clustering

---

Background & Objectives

Background

Apartment Listing serves as a comprehensive description of available apartment units, playing a crucial role in the real estate industry by aiding property owners, landlords, or real estate agents in attracting potential tenants or buyers.

Objectives

The project aims to enhance stakeholders' understanding of various apartments listed, utilizing machine learning techniques to analyze and predict apartment listing features and their impact on marketability and pricing.

Dataset Description

The dataset contains diverse information about apartments, covering the apartment's surroundings and the unit itself, along with offers on the apartment.

• Features: 19 Columns
• Entries: 12.2k Rows

Key Attributes:

• Unique ID for each apartment
• Location (longitude and latitude)
• District
• Proximity and grade of the nearest subway
• Number of rooms, floor, and material of the apartment
• Condition, category age, and price of the apartment

Tasks

Exploratory Data Analysis (EDA)

Exploration of the dataset to uncover relationships between building locations and prices, characteristics of buildings near subways, the correlation between building conditions and age, and the impact of district on offers.

Classification

Developed models to classify apartments based on features using Random Forest, K-Nearest Neighbour, and Neural Network. The models underwent parameter tuning with techniques like GridSearch and RandomSearch to optimize performance.

Regression

Focused on predicting apartment prices through regression analysis. Models such as Random Forest, XGBoost Regressor, and LGBM Regressor were evaluated based on metrics like R-squared and mean squared error, with parameter tuning to enhance accuracy.

Clustering

Clustering was performed to identify patterns or groups within the apartments based on geographical and feature-based similarities. Key features like district, proximity to the subway, building age, condition, material, and price were considered for clustering analysis.

Conclusion

This project applies various machine learning techniques to analyze and predict factors affecting apartment listings' marketability and pricing. It provides insights into the real estate market, assisting stakeholders in making informed decisions. For further information, please open Final Presentation.pdf (Indonesian)

Acknowledgements

We thank our mentors and peers for their guidance and support throughout this project.

---

Thank You for Your Interest in Our Apartment Listings Project!