Cross referencing the original issue

[nipunbatra]

nilmtk/nilmtk#153

This has been assigned to @RishiBaijal

[nipunbatra]

@RishiBaijal did some great work and reported some errors. I've copied them here.

1. Since our last conversation, I have downloaded the latest version of iAWE and I have tested all my code on that. The code that I have written is available on my forked repository.

2. Some of the functions that I have tested ran smoothly. I tried plotting the data and it works just fine. However, a lot of the functions showed problems. They are as follows:

   a. average_energy_per_period() : This function crashes spectacularly, and returns a mysterious message that says "ValueError: cannot reindex from a duplicate axis"

   b. correlation(): This function goes into an infinite loop when I try to find the correlation between the object and itself. I have no idea why this happens. Shouldn't it just return 1?

   c. dataframe_of_meters(): This function goes into an infinite loop for absolutely no reason whatsoever.

   d. energy_per_meter(): This gives a similarly frightening error to average_energy_per_second(). However, the interesting thing is this - if I try to run this function in the qtconsole, the kernel simply dies and then starts again.

   e. entropy_per_meter(): This throws "ValueError: total size of new array must be unchanged"

   f. fraction_per_meter(): Strangely, the error for this is very similar to the energy_per_meter() error.

3. I tried to write a loop that would go through all the functions of a particular class (the code for that is in the comments here). However, there are far too many nuances associated with each function to run it in a loop. Therefore, it seems like the only option is to run every single function manually. I know this seems like an ugly way of doing things, but I wasted quite a lot of time in trying to come up with something to do things dynamically inside a loop (doing it statically would take just as much time). The busy-work involved in both the cases seems to be the same. Therefore, I am going with the static technique in preference to the dynamic one.

4. As a result of the problems I have described above, I seem to have fallen behind schedule. I will need more time in order to complete this. There have been other problems. For example, the pairwise_correlation() function takes a lot of time to execute on the iAWE dataset.

What are the causes of the errors I have described above? Clearly it is not a problem with the dataset. Do you face the same problems on your machine? If there is something wrong with my code, I fail to understand what it is because I am simply calling the functions as they are. Is the usage wrong somehow? If so, what is the correct usage?

[nipunbatra]

I've made a quick start on this in the following commit f1445cf

@RishiBaijal: Let us wait for @JackKelly and @oliparson to do a quick review. After that you can write the code in the skeleton I've put.

[JackKelly]

I haven't yet looked at the code but... Thank you for identifying issues
with nilmtk. Whenever you find a bug in nilmtk, please can you submit an
issue report to nilmtk. One issue per bug. And please include a full stack
trace, and a minimal code example to reproduce the error. And please make
sure you are using the latest version of nilmtk as the code for the
functions you mention has changed recently.
On 14 Dec 2014 23:38, "Nipun Batra" [email protected] wrote:

I've made a quick start on this in the following commit f1445cf
f1445cf

@RishiBaijal https://github.com/RishiBaijal: Let us wait for @JackKelly
https://github.com/JackKelly and @oliparson
https://github.com/oliparson to do a quick review. After that you can
write the code in the skeleton I've put.

—
Reply to this email directly or view it on GitHub
#1 (comment).

[RishiBaijal]

Actually the text of the above comment has been pasted from an email that I had sent to @nipunreddevil earlier. That email contained hyperlinks to the code that I have written and the error messages. I will just put them here:

average_energy_per_period(): http://pastebin.com/CCX2b6Yc
energy_per_meter(): http://pastebin.com/dN32zVXX
entropy_per_meter(): http://pastebin.com/9xB44i8g
fraction_per_meter(): http://pastebin.com/iMgvhpuC

My script: https://github.com/RishiBaijal/nilmtk/blob/master/test_all_functions.py

I have opened a NILMTK issue here in order to address this particular bug. It can be found here: nilmtk/nilmtk#304. The issue contains exactly what I have written here.