maechler/nnilm

A reimplementation of Jack Kelly's rectangles neural network architecture based on Keras and the NILMToolkit.

nnilm's Issues

Sliding Window Approach for Online Energy Disaggregation Using Artificial Neural Networks

User 2to3.py in anacoda can translate this project

Hi maechler，
thanks for your code，I reproduce Kelly‘s result by 2to3.py in anacoda can translate this project and kelly's porject. In fact, that is a feasible way that use the tool to runing on python 3.x,

confusion about the choice of date range

Hi Maechler ,
Thanks for nice code. I am a new in nilm and have confusion about the choice of data range in NILM Area, could you please give me some advices if possible?

I want to compare algorithms, what is the standard for choosing date range? As I see in some paper, like Neural nilm, Sequence-to-point learning with neural networks for non-intrusive load monitoring, Sliding Window Approach for Online Energy Disaggregation Using Artificial Neural Networks etc, where author train and test specified houses in ukdale or redd without detail about begin-end(window) date.
If I want to compare my model with existing model described in paper, need i use same date range(windows) on same dataset?

Thank you so much,
Appreciate!

Question about std

Hi maechler，
Thank you for your code to help me complete my graduation project.I find in kelly's paper,target applicane's power divided by a hand-code 'maximum power demand',but in your code or kelly's response divided target's std .I am troubled by this.

What is the data you use to make predictions?

What is the wash_disher_eco?

tensorflow version problem

I am trying to install your project on my windows 8.1
I have a problem with tensorflow version as it is only supported for python 3.5 or more and your project need 2.7 and there is no tensorflow for such version on windows
Is there any way to overcome this problem?