RapidRate is a tool developed by CSIRO, using Artificial Intelligence techniques, that can quickly rate the energy efficiency of a dwelling using a relatively small number of inputs. RapidRate generates an estimated Star Rating that is aligned with the Nationwide House Energy Rating Scheme (NatHERS). It also generates estimated heating energy and estimated cooling energy.
RapidRate inputs include: dwelling type (house or apartment); floor area; external wall area by orientation; window area by orientation (including area double glazed); main wall, floor, and roof materials (including insulation); and postcode.
An Application Programming Interface (API) enables connection of RapidRate to any user interface or application.
RapidRate model accuracy
RapidRate development commenced in 2020 and has been an iterative process. RapidRate will continue to grow and evolve over time. The accuracy of the RapidRate model is influenced by several factors including the training data used, and the machine learning techniques used.
The current training data comes from the CSIRO NatHERS certificate database. This contains data for Australian dwellings built since June 2016. This means that the RapidRate model may not yet be fully representative of all Australian housing.
Some of the training data is used to train the model, and some is used to evaluate the model. Evaluation results reflect the ability of RapidRate to rate dwellings represented by the training data.
There are two ways in which the accuracy of the RapidRate model is described.
1. Prediction interval gives an idea of accuracy at an individual dwelling level
When a rating for an individual dwelling is conducted, point estimates for star rating, heating energy, and cooling energy are returned. In addition to this, prediction intervals for each measure are returned. For example, a dwelling may rate at 3.4 stars with 90% confidence that the value is between 2.9 and 3.9 stars.
2. Accuracy measures give an idea of accuracy of the model overall
A variety of measures are used to assess the accuracy of the RapidRate model.
Fig. 1 shows predicted star ratings against actual star ratings using the evaluation training data.
RapidRate model summary measures:
- R2 (test): 0.81
- Root Mean Square Error (test): 0.45 stars
- Mean Absolute Error (test): 0.31 stars
- Median Absolute Error (test): 0.2 stars
Working with CoreLogic
CSIRO are collaborating with property data analytics company CoreLogic to trial the use of RapidRate. See here for more information.
Last updated: 20/4/2023