Predicting user's activities is not an easy task. Specifically, the activities themselves are dependent to each other, and the information of activities are given in uncertain ways (i.e., it is hard to get fully correct information of events in reality).
Previously, I have tried Poisson processes to predict user's events, however it has been uncesseful model as the events are independents in Poisson process. Continuing researching for an appropirate model, I have had a good chat with Long about a model, still using Markov process. I think we came with an appealing model, which also works for dependent data. I am implementing the model with the data collected from the FigureEnergy system, however there is not enough information to conclude this model. For further testing, I will test it with a large appropriate data, which are released by MIT. Thanksfully the data can be obtained from Oli. I will try to implement this model asap for further analysis.
Then, I will have to define a scenario in a formal mathematical way, then an existing model will be chosen to implement as a benchmark for future comparison. In addition, the paper of "A model for Temporal Dependencies in Event Streams" will be needed to read and understand as well.
Thank you for the update Henry.
ReplyDelete> there is not enough information to conclude this model
Do you mean that the amount of data is not enough for training, or that the amount of information in the data is not enough? If so, what else is needed?
> Thanksfully the data can be obtained from Oli
Is this data not online?
I.e. do you need to wait for Oli to be back to go ahead with this?
> an existing model will be chosen
I am not sure I understand what you mean here.
Do you mean for example the model from the Kim et al paper or from the Gunawardana et al paper?
Thank you,
Enrico
I try to plot an emperical distribution of the frequency of the user's events. The longest data I have is 17 days of the user ecenergy22. Is it possible for you to pass me more data of user's labels in FigureEnergy.
ReplyDeleteI have asked Oli about the data he used for his work. Unfortunately, the data only shows the usage of the specific devices that have been observed by MIT team. No labels can be found in this data.
I have read the Gunawardana et al paper, the problem model they have tried to solve is the same problem that I am trying to solve. So, I probably focus to implement their model.
> Is it possible for you to pass me more data of
ReplyDelete> user's labels in FigureEnergy
Few weeks ago I sent you via email a link to a more recent snapshot of the data. Did you get that? There should be at least 4 weeks of data there, but probably (and hopefully) more.
Please send me an email if you do not find it.