As of the fourth quarter of 2024, the resale price index of residential units from the Housing Development Board (HDB) in Singapore was at 197.9, which means that HDB resale flat prices increased by 97.9 percent since the first quarter of 2009. The index tracks the overall price movement of the public residential market, compared to the base value from the first quarter of 2009, when the index value was equal to 100.

Dataset from Housing & Development Board. For more information, visit https://data.gov.sg/datasets/d_14f63e595975691e7c24a27ae4c07c79/view

Resale Price: Avg Valuation: HDB Flats: Bukit Batok: 5 Room Flat data was reported at 520,000.000 SGD in Sep 2018. This records an increase from the previous number of 500,000.000 SGD for Jun 2018. Resale Price: Avg Valuation: HDB Flats: Bukit Batok: 5 Room Flat data is updated quarterly, averaging 454,500.000 SGD from Sep 2002 (Median) to Sep 2018, with 54 observations. The data reached an all-time high of 640,000.000 SGD in Jun 2013 and a record low of 305,000.000 SGD in Sep 2005. Resale Price: Avg Valuation: HDB Flats: Bukit Batok: 5 Room Flat data remains active status in CEIC and is reported by Housing & Development Board. The data is categorized under Global Database’s Singapore – Table SG.EB027: Resale Flat Statistics.

Resale Price: Avg Valuation: HDB Flats: Geylang: 3 Room Flat data was reported at 270,000.000 SGD in Sep 2018. This records a decrease from the previous number of 305,300.000 SGD for Jun 2018. Resale Price: Avg Valuation: HDB Flats: Geylang: 3 Room Flat data is updated quarterly, averaging 268,000.000 SGD from Sep 2002 (Median) to Sep 2018, with 65 observations. The data reached an all-time high of 365,000.000 SGD in Sep 2014 and a record low of 115,100.000 SGD in Sep 2002. Resale Price: Avg Valuation: HDB Flats: Geylang: 3 Room Flat data remains active status in CEIC and is reported by Housing & Development Board. The data is categorized under Global Database’s Singapore – Table SG.EB027: Resale Flat Statistics.

Background

I've been sitting on this for quite a while and it's a project that I'm glad that I attempted because it forced me to learn new things through trial and error in order to enrich the dataset. Like mapping variables based on set/dict of values for the CPI, lease and GNI adjustment columns. Like trying to scrape Google Maps source code and getting blocked from their maps for a while before finding out how to use other geocoders responsibly. Like running multiple simulations of different linreg algorithms and analysing their variance contributions.

Anyway, this dataset can be used to create rich data visualisations or you can try using it for machine learning because of the large sample size in a small geographic area.

Dataset contents

ALL Prices 1990-2021 Mar.csv

The largest file by far, ALL Prices 1990-2021 Mar.csv, contains over 800k rows of transactions of Singapore Housing Development Board (HDB) resale flats. As the name implies, BTO, SERS, HUDC and private housing are not included, though resale DBSS flat transactions are treated as ordinary HDB flat transactions. Many of the columns in the file are calculated columns or mapped columns (based on supplementary information) like the CPI index and lease percentage columns. For the full metadata/glossary of how I derived these terms, see the bottom of this description and/or the file and column descriptions.

Balas Table.csv

Balas Table.csv contains the ratios of leasehold land value to freehold land value for each year of remaining lease, from 1 to 99. This table is used by Singapore Land Authority (SLA) in determining land valuations which affect property value since most land in Singapore is leasehold. As there are some mistakes/anomalies in the dataset with 100 and 101 years, I used the maximum values of 96% ratio when mapping in the 2nd version so please don't use the old version.

MAS Core Inflation.csv

This file contains SIngapore's core CPI index value for each month from January 1990 to February 2021 as compiled by MAS. For March 2021 and subsequent future transactions, you have to make estimates and also update this table based on new releases by MAS. For this dataset, I used March 2021=100.4.

complete.csv

Contains all UNIQUE BLOCK addresses, along with their geocoder-supplied full address (inconsistent, many missing) and more importantly, their latitude and longitude coordinates. As there are 9000+ addresses, they were first geocoded using a mix of Photon and Google Maps source code scraping (more accurate but doesn't give full address for quick checking). Then, I manually looked through the addresses and coordinates to find and update all blatantly wrong (outside Singapore or wrong neighbourhood) and most slightly inaccurate (correct neighbourhood and/or street but tagged to wrong block number) for a total of around 1600 addresses, many in Whampoa/Boon Keng, Sengkang, Yishun and Woodlands.

gni per capita.csv

As the name implies, this file contains Singapore's GNI per capita in nominal S$ for the years 1990-2020. For 2021, you have to make an estimate based on the projected economic recovery from COVID-19 until the actual value is released. For this dataset, I used 2021=75000.

HDB machine learning.xlsx

This file is my own basic analysis of (numerical) variables that potentially help to determine the final resale price (measured by inflation-adjusted price per square metre). I used 5 linear regression algorithms and tested each variable individually as well as tried to maximise the R^2 using multiple linear regression with as many relevant variables as possible. I also included the correlation matrix between all the variables and that for relevant variables which helps in calculating the incremental contribution to variance for each variable.

Usefulness in training models?

As i've shown in the "HDB machine learning.xlsx" workbook, some variables are more influential than others but even the amount of variance contributed changes depending on the conditions applied. Various multiple linear regression models i've tried can only post up to 0.60+ in combined R^2, which means that up to 40% of variance in the inflation-adjusted price per square metre/foot flat prices is essentially just random noise or could have another hidden variable! Perhaps you can try to find another strongly related variable? Some ideas are proximity to MRTs/bus stops, ratio of HDB to private housing, average household size, other housing to population ratio indicators? It's important to consider whether correlated factors are causes or effects as well.

More importantly, do you think you can train a model to post much better numbers than multiple linear regression?

Metadata for ALL Prices file

Original columns: `month, town, flat_type, block, street_name, storey_range, area_sqm, flat_model...

Housing Index in Singapore increased to 210.70 points in the first quarter of 2025 from 209.40 points in the fourth quarter of 2024. This dataset provides the latest reported value for - Singapore Property Price Index - plus previous releases, historical high and low, short-term forecast and long-term prediction, economic calendar, survey consensus and news.

This analysis presents a rigorous exploration of financial data, incorporating a diverse range of statistical features. By providing a robust foundation, it facilitates advanced research and innovative modeling techniques within the field of finance.

HDFC Bank: (HDB) Riding the Wave of India's Growth

Financial data:

• Historical daily stock prices (open, high, low, close, volume)

• Fundamental data (e.g., market capitalization, price to earnings P/E ratio, dividend yield, earnings per share EPS, price to earnings growth, debt-to-equity ratio, price-to-book ratio, current ratio, free cash flow, projected earnings growth, return on equity, dividend payout ratio, price to sales ratio, credit rating)

• Technical indicators (e.g., moving averages, RSI, MACD, average directional index, aroon oscillator, stochastic oscillator, on-balance volume, accumulation/distribution A/D line, parabolic SAR indicator, bollinger bands indicators, fibonacci, williams percent range, commodity channel index)

Machine learning features:

• Feature engineering based on financial data and technical indicators

• Sentiment analysis data from social media and news articles

• Macroeconomic data (e.g., GDP, unemployment rate, interest rates, consumer spending, building permits, consumer confidence, inflation, producer price index, money supply, home sales, retail sales, bond yields)

Potential Applications:

• Stock price prediction

• Portfolio optimization

• Algorithmic trading

• Market sentiment analysis

• Risk management

Use Cases:

• Researchers investigating the effectiveness of machine learning in stock market prediction

• Analysts developing quantitative trading Buy/Sell strategies

• Individuals interested in building their own stock market prediction models

• Students learning about machine learning and financial applications

Additional Notes:

• The dataset may include different levels of granularity (e.g., daily, hourly)

• Data cleaning and preprocessing are essential before model training

• Regular updates are recommended to maintain the accuracy and relevance of the data