HAJBL Consulting
Machine learning consulting
As well as our own apps, HAJBL Pty Ltd builds machine learning systems for organisations that want to get more out of their data. Most of the work involves models that predict, value or forecast something, and we test each one carefully against simpler approaches before anyone relies on it.
What we do
- Predictive modellingWe build models that forecast the outcomes your decisions depend on, and we backtest them against simple baselines so you can see whether they actually add anything.
- Valuation enginesThese systems keep a running market value for a set of assets as prices and other inputs change. The assumptions are visible, and you can adjust them.
- Demand forecastingWe forecast demand for products, services or locations, and each forecast comes with a likely range as well as the central estimate.
- Signals and decision supportWe produce ranked signals that help a team decide where to put its effort next, with an explanation of what drives each one.
- Data pipelines from public dataWe turn open government, statistical and market data into clean datasets that can be rebuilt from scratch whenever the source data changes.
- Model reviewsWe review models that are already in use, looking at the data, how they were validated, any leakage, how uncertainty is handled and whether the results hold up.
Selected work
Here are a few examples of the systems we have built. Client details are kept confidential.
- Mining valuationA platform that values listed mining projects at current market prices using their published technical studies.
- Regional prediction engineA prediction engine for outcomes in Australian local government and regional areas.
- Local business demand predictionModels that forecast demand for local businesses.
We also build systematic signals for financial markets.
How we work
- We start with the decision the model needs to support and work back from there to the data.
- Every model is compared with simple baselines, and we report how uncertain its results are.
- We write down the assumptions and limits of each model so that its results are only used for what they can support.
Talk to us
Tell us what decision you are trying to make and what data you have, and we will give you a straight answer about whether machine learning can help.