search

MingjunZhong / NeuralNetNilm

Sequence-to-point learning for non-intrusive load monitoring (energy disaggregation)
71 stars 21 forks source link

'.npy' file #1

Open moondalll opened 5 years ago

moondalll commented 5 years ago

Hello, MingjunZhong! I'm GyuMoonKyung in Korean Graduate School student. I'll contact you for more research on energy decomposition technology using your paper. I'm trying to follow your coding practices, and having difficulty processing input data on the way. I was using '.npy' files in your coding, and I wonder how you got those files. If you have converted the file, please teach me. I look forward to your kind reply. Thank you, God bless you.

MingjunZhong commented 5 years ago

@moondalll : .npy file is a file generated by using numpy. You can read .npy file using numpy methods. Basically, .npy files were converted from .csv in our examples. You can have a look at this: https://www.numpy.org/devdocs/reference/generated/numpy.lib.format.html Hope it is helpful. Mingjun

moondalll commented 5 years ago

Thanks a lot! I appreciate your kindness, I'll do my best! God bless you. -----Original Message----- From: "Mingjun Zhong"notifications@github.com To: "MingjunZhong/NeuralNetNilm"NeuralNetNilm@noreply.github.com; Cc: "moondalll"byulldall@naver.com; "Mention"mention@noreply.github.com; Sent: 2018-11-22 (목) 23:27:34 Subject: Re: [MingjunZhong/NeuralNetNilm] '.npy' file (#1)

@moondalll : .npy file is a file generated by using numpy. You can read .npy file using numpy methods. Basically, .npy files were converted from .csv in our examples. You can have a look at this: https://www.numpy.org/devdocs/reference/generated/numpy.lib.format.html Hope it is helpful. Mingjun — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

2ndMessiah commented 5 years ago

Hi Mingjun, thanks a lot for providing the source code here. I have confusion about the data format requirements of npy files. Would you brief related information of ndarray shape,required physical quantity, sample frequency, etc? Much obliged.

MingjunZhong commented 5 years ago

Hi @2ndMessiah, I have explained the *.npy files in previous comments. The npy file used in our code is a 1-d numpy array. It is the time series for the mains (x) and the appliances (y) correspondingly. So in the code these data are fed into the DataProvider.py to generate the training/test data sets - pairs of windows for x and y. So in the npy file, there is no sample frequency recorded, which is essentially the time stamps. I think the time stamps should be in the csv file. I am not sure about your question about physical quantity. Any questions please ask. Mingjun

aleonnet commented 5 years ago

Hi @MingjunZhong I've read your article and would like to reproduce your results. Would you share the training data you've used? Or the code to extract it from ukdale dataset.

   default='../data/uk-dale/trainingdata/small/', 
   help='this is the directory of the training samples')

Thank you.