HACKER Q&A
📣 gwnywg

Is there any real value in applying ML to process time series data?


Hi,

I work a lot on images processing and extracting insights from it. There seems to be demand for this kind of service. My university background is closer to time series data, i.e. digital signals processing and modelling, I was wondering if same methods can be applied here. In industrial processes it makes sense to try to model various parts of the process to be able to predict possible fault. This is cool and complicated, and also there are no industrial plants in my environment to go and explore real life data to work with.

The other application of modelling based on time series is within financial markets speculation (forex, stock exchange etc.), though I am not partucularly interested in this area, it appears to me a bit like gambling and as such not something I would like to be involved in.

So I wonder what other use cases for time series modelling are. I have two ideas to build something that may turn out to be useful.

1. I was reading it is possible to find correlations between time series streams, and this could maybe lead to a model that suggests to turn some equipment off because it is not required (based on availability of electrical consumption data)

2. Processing of audio streams and building generative model to create some interesting sound effects (not necessarily music); or the other way around, to detect some events based on sound (i.e. car accident)

Do you know interesting projects that apply machine learning methods to process time series data to do something useful with it? Or maybe you wish a project existed for some particular problem you feel is worth solving?

  👤 ramtatatam Accepted Answer ✓
I was working for a company collecting readings from IoT sensors of all sorts (anything between temperature/humidity to weirdest things like weight of poultry). One issue we had was that we had to manually group related readings, this was specifically painful for data collected from bigger data sources with flat structure, like one sensor box with many wires connected into it and at the end of each wire there were multiple sensors collecting data - those sensors were then represented in the flat structure and only correct naming was allowing us to figure out what sensors were related to each other. I don't know if it is possible to group related streams of data purely based on values, I would be very interested to learn what is possible here. Generally data interoperability is the big problem in this field, manufacturers (and engineers) naming their things differently. One sensor manufacturer is calling temperature `Temp`, the other manufacturer is calling it `temperature`, or just `T1`... And I don't know if this problem is big enough, we were simply figuring this out each time we had new player introduced to us.