Hyper-automated predictive analytics
Updated: Apr 28, 2021
Hyper-automated predictive analytics, or self building forecast models, in plainer English, are the latest thing in the world of business analytics - but are they for you?
Advances in machine learning have enabled computers to autonomously build and optimise time series models based on the data they are fed regarding predictor and target variables.
For example if you wanted to forecast the output of a solar farm tomorrow you might take predictors as levels of insolation, sunrise and sunset times, and temperature (since it may affect the efficiency of the solar panels). You would take historical data for these and get the machine to calculate the best algorithm to correctly predict the past values for electricity output per half hour. The machine would do this by creating a first-cut model and then try variants to see if it could improve on accuracy. It would then optimise the algorithm to deliver maximum accuracy.
In olden days, and by that I mean less than a decade ago, you would use sophisticated modelling tools deployed by a team of data scientists. The whole process would typically take a few months.
Today, the data scientists define the problem - and the machine builds the model and optimises it, in one pass through the data in a matter of seconds. Once you have access to the engine to build the models - each model build and model run costs virtually nothing since programmers and developers are not needed to write the models. They were needed to write the model building engine.
All fascinating from a technology perspective but what does it mean for you and your organisation?
It means that adopting hyper-automated predictive analytics can lead to significant performance improvement
This is because good data can create good models which can create good forecasts
and good forecasts can help you to make better decisions,
and better decisions can lead to significantly improved performance
Could you have access to the data which has within it the intelligence of how the world works?
Would having more forecasts, or having better forecasts make a significant change to your decision making?
Would alternative decisions have a material effect on your performance?
If the answers to all of the above is yes - then you definitely need to be using hyper-automated predictive analytics.
We have pulled together a 5 minute quiz to help you decide on whether you should be taking a closer look at hyper-automated predictive analytics.
Whilst the quiz is really easy to complete, it does assume you have a reasonable understanding of forecasting and modelling in your organisation. If you are interested in the subject, but feel you are not as close to the subject matter as you could be why not complete the quiz with a colleague who is.
You don't need to sign up to take the quiz and get the answer - however if you do give us your email we can then send you the rationale behind each of the answers which I think you will find illuminating.