DataFrame.diff/pct_change

[joshuakoh1]

There's no difference function?

https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.diff.html
https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.pct_change.html

[joshuakoh1]

@risenW if this isn't a near term priority could you point me in the direction of how to hack together a difference operation? Which functions should I chain?

Thanks!

[risenW]

@risenW if this isn't a near term priority could you point me in the direction of how to hack together a difference operation? Which functions should I chain?

Thanks!

The might help https://ourcodeworld.com/articles/read/557/how-to-calculate-a-percentage-change-increase-and-decrease-from-2-values-in-javascript

[joshuakoh1]

@risenW I mean how do I pass n, n-1 into the same function across the frame? I couldn't find anything that accesses the previous row

[risenW]

@risenW I mean how do I pass n, n-1 into the same function across the frame? I couldn't find anything that accesses the previous row

You need to write a custom function and then apply it to the DataFrame.
You can use the apply function to apply custom functions to rows or columns of a DataFrame. See https://danfo.jsdata.org/api-reference/dataframe/danfo.dataframe.apply.

[joshuakoh1]

@risenW I'm not sure how to access the previous row. When I apply, I only get the current row but I would like a difference of the row above. df.apply only passes the same row.

keys = [A,B,C]
cols = [[1,2,3],[4,5,6],[7,8,9]]

I would like to diff B/C e.g. (5-2)/(6-3) then (8-5)/(9-6)

[NeonSpork]

If nobody else is actively working on implementing those two functions, I can try my hand at making the equivalent function in danfo? :)

I may need a guiding hand writing tests however, though I'll do my best!

[risenW]

If nobody else is actively working on implementing those two functions, I can try my hand at making the equivalent function in danfo? :)

I may need a guiding hand writing tests however, though I'll do my best!

PR is definitely welcome! You can read the contributor guide here first: https://danfo.jsdata.org/contributing-guide If you need any help, just ping me in the PR

[NeonSpork]

I started digging into pandas, and essentially followed a rabbit hole further down into numpy. So this will likely take me a while, as there is a lot going on in the diff function!

It would appear that the best method of implementing the diff function is to emulate the numpy function which uses slice to isolate the relevant row/column and calculate the difference the python API for tensorflow essentially took that same numpy logic too.

Following that logic I can leverage the TensorflowJS method tf.slice() to achieve something similar. I've cloned and branched off, and will let you all know of any speedbumps or hiccups! Let me know if you have any thoughts or advice @risenW / @steveoni 🚀

[joshuakoh1]

@NeonSpork thanks for the work. @risenW, any idea when is the next version bump featuring this?