Project Description
Property buyers often have to use various factors to determine which property to buy/invest in. There is also no certainty if the price of the property will increase in the next few years.
Objectives
- Come up with a model to predict property prices, backed by data
- Identifying factors that affects the prices of property
- Predict future prices accurately, up to 10 years
Objectives
- Identifying Variables: PSF, PPI, Volume, MRT, Schools, Age, Covid
- webscraping.ipyn: webscrape property transaction data from Edgeprop
- MRT and school scrape.ipyn, onemap scraper.ipyn: webscrape MRT and school data from OneMap
Data Cleaning
ModelDB grp by.xlsx in Raw Data folder
Modelling
Iterative process looping through:
- Run linear regression on possible models
- Extract best model (using adjusted r^2)
- Examine residual plots
- Investigate variations of variables
Results
Assumptions
- Near MRT" defined to be within 350m from any station
- Transactions between 2020 and 2022 labelled as Covid
- Tao Nan School and Kong Hwa School classified as "Good" schools in the region
- Latest average PSF assumed for quarters without transactions
- Average PPI growth rate is assumed constant when forecasting
- Non-drastic change in behaviour by seller
Limitations
- Model specific to District 15 & 99-year leasehold
- Dataset limited by PPI being published quarterly; data had to be grouped quarterly
- Unit specific variables (i.e. bedroom size, level, window direction) not considered due to nature of data grouping
Tools Used
Python
Pandas
Selenium
R
HTML
GIT
Web Scraping
Math Modelling
Consulting