MSc. Information System Management
Kyaw Khine Soe (3026039)
Data Mining and Business Analytics
Boston Housing Dataset Analysis.

Table of Contents Introduction 3 Problem Statement 3 The associated data of Boston 5 Data pre-processing / Data preparation 8 Clustering Analysis 11 Cluster segment profile 17 Regression Analysis 18 Predictive analysis using neural network node 19 Decision tree node 21 Regression node analysis 23 Model Comparison 24 The recommendation and conclusion 26 Bibliography 27

Introduction

This report included part of assignment for the Data Mining and Business Analytics. This report based on the Boston Housing Dataset to describe prediction, cluster analysis, neural networks and decision tree nodes. Boston Housing is a real estate related dataset from Boston Massachusetts. This is small dataset with 506 rows can show prediction of housing price and regressing using decision trees and neural networks over this dataset. This report shows analysis of the property price over the size, age of property, environment factor such as crime rate, near the river dummy, distanced to employment centers and pollution.
Problem Statement
In relation to housing intelligence, real estate are usually concerned with following common business concerns: 1. Which area are high rates of crime? How crimes rates effected on housing price?
How can reduce the crime? 2. Which area is most/lease house price base on rooms in house/ area and pollution? What are the characteristics of them? 3. Does people willing to pay for more cleaning air? Does housing price near river chase is high or near industry zone? 4. How the ratio of pupil and teacher effect on the society? How is it effect on the crime rate of town? 5. How minorities group effect to the housing price? Are they related to crime rate? 6. What are the house