Predicting customer behaviour offers major benefit to any company and can help increase or sustain market share. However, the data is often very large and creating algorithms and using software that are scalable is often a challenge.
NLE is positioning themselves as a multi-utility provider and are moving towards being service based. The Head of Data Science at the NLE wanted management to have insight into the energy usage of their customers and then to provide actionable insights based on this behaviour.
The Head of Data Science presents his findings on how MATLAB can be leveraged to perform large data analysis on the cloud. In order to be successful, they needed to analyse and predict customer behaviour so that they could provide an accurate service to their customers, such as heating control via geofencing.
NLE chose MATLAB for their algorithm development because of it is scalable, well documented and it integrates well with other software. By using MATLAB and Hadoop they managed to scale their algorithms to first work locally on their machines and then scale up to a cloud service. Faster development iterations enabled them to find and solve problems quickly and provide management with answers in minutes instead of hours.
By using MATLAB on their machines NLE developers could test algorithms quickly and then push the algorithms up to the cloud to process the large volumes of data. The increase in performance and scalability allowed them to save on costly processing time and reduce the need for expensive in-house clusters.
MATLAB’s inherent performance capabilities were of great benefit to NLE as recognised by the Head of Data Science: “[one of the lessons learnt was] only use Spark when you really need it, MATLAB out of the box performance is enough for most usage”.