Using nilmtk on a private dataset

[redouane-b]

I am interested in using the nilmtk library to analyze a private dataset. However, I am unsure about how to format my dataset to work with nilmtk. Specifically, I am wondering:

• Is it possible to use nilmtk on a private dataset?
• If so, what is a good format that can be used to turn my pandas dataframe into a nilmtk dataset?

I have looked through the nilmtk documentation and searched for similar issues on Github, but I couldn't find any clear guidance on how to proceed. Any help or suggestions would be greatly appreciated. Thanks!

[crn565]

If possible. The easiest way is to create a new converter from csv measurement files for example based on the IAWE dataset. If you don't have the csv files, you can extract them in python with a few steps. In my case I have created several converters for OZM hardware and also for a new open hardware OMPM. If you want in github/crn565 you can find all the code

[crn565]

In this repository https://github.com/crn565/DSUAL in addition to describing the new datasets created with open hardware, it describes how the new converters are made to generate the datasets to be used in NILMTK.

[PeterQuinn925]

Similar situation - looking for pointers

Hi,

I'm a homeowner with a power meter. I'm storing the power data timeseries once a second into a database. I want to disaggregate the various appliances without instrumenting each of them. I've been reading through and experimenting with the various NILMTK data sets and algorithms. It seems all the datasets have appliance load data in them.

I have no trouble writing python or manipulating the time series data in a dataframe. But I'm stuck trying to figure out which algorithm might work in my application and how to modify one of the examples to my data set.

Has someone done this before? Is there a site with some samples? Or a tutorial on how to do it myself? Or even suggestions on which papers to read?

More details on my project here: https://hackaday.io/project/196939-house-electricity-monitor