Article By Fiona Morris
Thursday, November 14, 2019

Backed By Data

Forecasting by the TUDI platform is based on analysing data and then visualising the insights into 5 Indicators in the TUDI HotSpot™. The most important of these indicators is ‘Buyer Demand’, and an important first step for TUDI was to validate the correlation between the Buyer Demand indicator and Median Sale Price.

Buyer Demand is the most important of the
5 Indicators used to determine a TUDI HotSpot™ Suburb

The Buyer Demand Indicator recognises suburbs where demand is about to or is exceeding supply. It also assesses how quickly demand has been increasing compared to supply based on historical data. So it is more sophisticated and relevant than just a straight measure of buyer demand.

Expert analysis

TUDI has teamed up with the University of Wollongong (UoW) SMART Infrastructure Facility, which has a dedicated team of researchers who specialise in data analytics, modelling and economics. Together they’ve been studying the evolution of the housing market in Australia and the ability to forecast suburb growth.

In addition, researchers have been using advanced data models and artificial intelligence (AI) to validate the accuracy of  TUDI’s proprietary technology, in particular, the Buyer Demand indicator.

The good news is that the evidence from this research demonstrates a very strong correlation between TUDI’s Buyer Demand indicator and suburbs’ price growth.

Deep-dive into the science

Once the correlation between Tudi’s Buyer demand indicator and a suburb’s price growth established, wouldn’t it be great if you could forecast the Buyer Demand indicator, and subsequently suburb growth (or fall)?

TUDI uses two different forecasting models to predict Buyer Demand. The first is known as ARIMA (autoregressive integrated moving average), a statistical method that is a good predictor for continuing trends.

And the second is LSTM (long short-term memory). LSTM is a machine learning type method feeding data through a neural network giving predictions based on what has been observed in the data training set.

Historical property price data from the past five years is fed into each of these forecasting models, which are trained to interpret the datasets of specific suburbs to predict the Buyer Demand indicator for 5000 suburbs in Australia.

Using the predicted Buyer Demand results from ARIMA and LSTM, TUDI then analyses them with the MAPE (mean absolute percentage error) formula. MAPE is a measure of prediction accuracy of a forecasting method in statistics – it provides a percentage value of how accurate the data is. 

ARIMA and LSTM will forecast varying results, and by running them through the MAPE formula, it can be shown which forecasting model is producing the most accurate results for each particular suburb at any one time.

The accuracy of forecasting the Buyer Demand indicator

TUDI has shown its Buyer Demand  prediction accuracy to sit very high at:

1 month – 84%
6 months – 71%
12 months – 63%

The proprietary AI technology is constantly optimised as more data for each suburb becomes available. This will, over time, generate even greater accuracy in its property price predictions.

Property investors signed up to TUDI’s Investor Plan have access to the ‘Buyer Demand’ data on over 5,000 suburbs Australia-wide. And with such high suburb price prediction accuracy, you can be sure the insights are providing the best intel for you to make smart property investment decisions.

Examples of the 5 Indicators in the TUDI HotSpot™ across various suburbs.