I'm currently doing my graduate research for my Master's degree at a particular firm. Initially, I was supposed to classify images of smart meters in homes using deep learning technology. Unfortunately, the dataset available is extremely limited, so I'm not able to do this particular project.
However, there are some datasets available of smart meter readings. Can you guys tell me what I could potentially retrieve from these kind of datasets? It must be within the field of deep learning or machine learning. I prefer deep learning over machine learning. The supervisor from the firm suggests finding outliers and predicting faulty equipment this way. But I'm afraid my supervisor from the university Isn't going to accept it, due to the simplicity of the project and because it leans too much towards data analysis, rather than data science.
The data can be very basic:
Time Usage 2014-11-28 00:30:00 2.085885556 2014-11-28 00:45:00 2.391335556 2014-11-28 01:00:00 2.092666667
But also more comprehensive like the Kaggle dataset smart meters in London
On the one hand, I have to do a research within the field of data science, but on the other hand, I have to deliver a useful end product/application for the firm I'm doing my internship at. I'm having a hard time finding a balance between these two.
Looking forward to your suggestions.