|
Aaron Meurer
@
asmeurer
ABQ
|
|
Math. Python. @SymPy. Working at @quansightai.
|
|
|
9.866
Tweetovi
|
505
Pratim
|
1.557
Osobe koje vas prate
|
| Tweetovi |
|
Aaron Meurer
@asmeurer
|
12 h |
|
It's good to see that the word "webmaster" is still alive and well in 2020.
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
12 h |
|
Webmaster of my own domain pic.twitter.com/2yYsOxA2ot
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
4. velj |
|
Probably a @pandas_dev dev could shed some light on this?
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
3. velj |
|
Emacs tip: Add (global-visual-line-mode t) to your .emacs to make it the soft wrapping respect word boundaries.
This is useful if you are editing a document where you cannot hard wrap lines (auto-fill-mode disabled), or if your split frames are narrower than your fill column.
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
3. velj |
|
I use pandas very rarely so I was going by what it looks like rather than experience. I expected "index" to mean they are the indices. If it works that way for arrays that does indeed seem inconsistent.
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
3. velj |
|
I expected an error but then when I saw your answer I remembered pandas has missing value support. What did you expect?
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
3. velj |
|
"The line, obtained from a linear regression, is statistically indistinguishable from identity (y = x)."
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
1. velj |
|
The best way to get people to test a release is to make the release.
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
1. velj |
|
With ctypes you have no static type checking (and bugs often mean segfault). OTOH it's effectively "pure Python" (no build or platform specific packages).
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
31. sij |
|
You understand that it improves readability but you still avoid it?
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
31. sij |
|
I see that as ungrammatical. An equation should be treated like it's part of the prose. See rule 3 here home.sandiego.edu/~severn/p480w/…
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
31. sij |
|
I just do this because it's a good habit. If you have a display mode equation, the comma or period has to go inside the math environment.
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
31. sij |
|
Why is it @me instead of just me? You really going to let one dude parking on the me username impede your user experience like this?
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
30. sij |
|
A professor at my university who taught logic lectured in the same way (copying the paragraphs from his hand written notes to the blackboard word-for-word). He was one of the best professors I've had.
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
30. sij |
|
Something that hasn't really been mentioned about Python 2: it's the last version of Python to remain supported for a longer time period. Organizations are going to have to start upgrading their Python versions on a regular basis now.
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
29. sij |
|
... But there isn't always a clear dividing line between something that doesn't need to be reusable and something that does. Often things cross the threshold and it's nontrivial to either transfer code out of a notebook or make a notebook reusable.
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
29. sij |
|
To me the main advantage of notebooks as a reusable file format is it removes a lot of boilerplate from certain kinds of data analysis workflows, at the cost or reusablity...
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
29. sij |
|
If the main strength of the notebook is interactive computing, then why is it designed around a file format? It feels like most of the purported advantages of "correct" usage of notebooks are also there with interactive REPLs like ipython, with none of the disadvantages.
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
29. sij |
|
That seems like itself based on looking at negligible probabilities in the tail. The probability of the average of Gaussian data being dominated by a single new datapoint becomes astronomically improbable after you have a certain amount of data.
|
||
|
|
||
|
Aaron Meurer
@asmeurer
|
29. sij |
|
My point is you have a sample of some data which follows some probability distribution. That distribution is such that the number of datapoints needed to naively estimate the mean far exceeds the number of points available.
|
||
|
|
||