NLE Predicts Customer Behavior Using Big Data Analytics with MATLAB in the Cloud
Rachid el Mimouni, NLE (formerly Nederlandse Energie Maatschappij)
NLE collects massive amounts of business data to predict customer behavior using big data analytics with MATLAB® in the cloud. As a result of changing its process, the company reduced calculation times from five to eight hours to 10 to 15 minutes.
Published: 25 Mar 2021
The main thing we were doing in this project was trying to predict customer behavior using big data with MATLAB. Telling you that you've used, let's say, 100 kilowatt hours yesterday, you say, OK, doesn't mean anything to me. What makes more sense to you is we would expect you to use 100, but you used 140. That's the reason we said, OK, it's time to take the next step, and we started adding a smart thermostat, which is basically an IoT device to give people insight in their energy usage.
And by using data, we are capable to predict what actions lead to what behavior, predictions on the level of a person. Making prediction on these levels also means more data because you are propagating the data for every customer. In order to do so, we needed maximum parallelization. But I cannot go to my boss and ask him for 500 computers. He'll probably say, are you crazy? That's also what he said.
So your solution is the cloud, the possibility to have those 500 computers but not the burden-- and with burden I mean the costs-- of those 500 computers. So there's no upfront investment. The result eventually is that we are running in 10, 15 minutes instead of 5 to 8 hours, and which also means that if the management asked me to run another scenario, I'm like, OK, give me 10 minutes. You'll have another scenario on your desk.