I have a time series recorded data of temperature. This is what my data looks like:
The change in data represents specific event or a class which I would like to detect when new incoming data. Because I cannot label the data manually I cannot do supervised learning. So, I thought take this sample data and then do some clustering to automatically make 5 classes(clusters). But I am not sure how this will be applied to the new incoming data?
I am naive in time series or data analysis (so not aware of many terms). I can use the sklearn or statsmodels library in python, but cannot use the LSTM or any Neural Network techniques. I searched about the auto correlation and drew the ACF graph of my data which looks like this:
I am not aware what this graph is saying? And not sure what to do with the auto correlation in my scenario. One more thing I had read is the HMM which may be helpful, but again I have zero knowledge on how to apply it or which route to go for this simple problem.
This is what I would like to do:
The data is simply the values taken from sensor and I would like to find out how many types different changes are present in this data.
After that, I would like to figure out which kind of change the new incoming data fall into.
