stable creek
Soooo I just installed a bunch of external Python packages
I just did: pip install package name
In my windows commands prompt
I installed things like numpy, Pandas, scikit, statsmodels, matplotlib, seaborn, requests, and xgboost
Idk if what I did was bad or okay. I'm new to this stuff.
I also don't know how to use them. Do I import the package? Notnd it's global or local or whatever
uneven sirenBOT
Soooo I just installed a bunch of external Python packages
I just did: pip inst...
@stable creek
Python help channel opened
azure flicker
!d pip
uneven sirenBOT
azure flicker
That's essentially what you need to do: pip install package_name...
To use a module you go: import package_name in your Python code.
Note that you might have multiple Python installs, and multiple environments using those Pythons. You need to install into whichever environment you're using - they all have distinct sets of packages.
For this reason we recommend going:
py -m pip install package_name...
using the python from the environment for py.
stable creek
So what's wrong with having multiple environments using them?
azure flicker
Importing a package, or some name from a package, is essentially an assignment statement:
import numpy as np
loads the numpy package (if not already loaded) and bind the name np to the numpy package. From then on you use the name np to access things from inside numpy.
Normally the imports are the first statements in your script/module, and so the names they bind are globals.
But like any other assignment, if you do them inside eg a function, they're locals.
azure flicker
So what's wrong with having multiple environments using them?
Nothing. But to use a package in an environment you need to install it for that environment. There's no "install for everywhere".
stable creek
Global to the Python session, global to the IDE I'm using (IDLE or jupyter), or global to my computer?
So I have to use import for each script where im using it?
storm coyote
Global to the Python session, global to the IDE I'm using (IDLE or jupyter), or ...
all 3
azure flicker
If a script uses a name from some package you need to import that package in the script. In order to have the package available as a name, which is the way you get to use what's in it.
storm coyote
So I have to use import for each script where im using it?
yes
azure flicker
Global to the Python session, global to the IDE I'm using (IDLE or jupyter), or ...
A package install is associated with an environment. So your pip install package_name installed in the the default Python install on your system.
Often an IDE makes an environment derived from the default Python (or another, if you do some extra stuff). You will need to install packages int that environment.
stable creek
I just checked on my idle and got this
i know you guys prefer '''py but like
its a massive block of text
storm coyote
you
you're in the python REPL, which is basically a live python programming enviroment
twin rivet
yes
technically not true if you import * but i suppose that is frowned upon and i regret even saying this
stable creek
uhh
storm coyote
technically not true if you `import *` but i suppose that is frowned upon and i ...
that wouldnt work
stable creek
you're in the python REPL, which is basically a live python programming envirome...
meaning that the packages are...?
storm coyote
!e ```py
import *
:x: Your 3.12 eval job has completed with return code 1.
001 | File "/home/main.py", line 1
002 | import *
003 | ^
004 | SyntaxError: invalid syntax
i mean from something import * of course
storm coyote
i mean `from something import *` of course
that's very different
storm coyote
meaning that the packages are...?
everything you're seeing there comes built in with python
twin rivet
so, a good way to show him the difference here may be to have him set up a venv?
storm coyote
so, a good way to show him the difference here may be to have him set up a venv?
no
So I have to use import for each script where im using it?
they still have to import things, star imports are still importing things
twin rivet
i mean to see the differences in what packages are installed between environments
stable creek
so uh
storm coyote
i mean to see the differences in what packages are installed between environment...
that's not what you originally said, anyways
stable creek
this didnt work
import numpy as np
nums = [1, 2, 3, 4, 5]
mean(nums)
there can be several reasons why this is wronng
but
idk what
again, new to coding
azure flicker
Well, where's mean supposed to come from?
storm coyote
again, new to coding
what was the error?
stable creek
Well, where's `mean` supposed to come from?
the numpy package? 🥺
storm coyote
the numpy package? 🥺
!d numpy.means
uneven sirenBOT
stable creek
NameError: name 'mean' is not defined
Ah. So if you mean this: https://numpy.org/doc/stable/reference/generated/numpy.mean.html
you want to say np.mean.
storm coyote
!d numpy.mean
uneven sirenBOT
azure flicker
This is why the import: to give you a name via which to access the functions etc from a package.
So you've imported numpy as the name np. Then np.mean gets the mean function from the numpy package.
azure flicker
The name np is a common convention, also pd for pandas. To reduce your typing burden.
stable creek
linear regression is a base function
azure flicker
Well Python has a heap of bulitin functions.
stable creek
only like 40, no?
storm coyote
meanwhile in R mean is already a base function 😭
!d statitics.mean
uneven sirenBOT
storm coyote
!d statistics.mean
we have it built in too, but you still have to import it
uneven sirenBOT
twin rivet
well, don't confuse "built in" functions with standard library
azure flicker
!e
import builtins
print(dir(builtins))
stable creek
i know min, max, sum, and abs are base
uneven sirenBOT
!e
import builtins
print(dir(builtins))
:white_check_mark: Your 3.12 eval job has completed with return code 0.
['ArithmeticError', 'AssertionError', 'AttributeError', 'BaseException', 'BaseExceptionGroup', 'BlockingIOError', 'BrokenPipeError', 'BufferError', 'BytesWarning', 'ChildProcessError', 'ConnectionAbortedError', 'ConnectionError', 'ConnectionRefusedError', 'ConnectionResetError', 'DeprecationWarning', 'EOFError', 'Ellipsis', 'EncodingWarning', 'EnvironmentError', 'Exception', 'ExceptionGroup', 'False', 'FileExistsError', 'FileNotFoundError', 'FloatingPointError', 'FutureWarning', 'GeneratorExit', 'IOError', 'ImportError', 'ImportWarning', 'IndentationError', 'IndexError', 'InterruptedError', 'IsADirectoryError', 'KeyError', 'KeyboardInterrupt', 'LookupError', 'MemoryError', 'ModuleNotFoundError', 'NameError', 'None', 'NotADirectoryError', 'NotImplemented', 'NotImplementedError', 'OSError', 'OverflowError', 'PendingDeprecationWarning', 'PermissionError', 'ProcessLookupError', 'RecursionError', 'ReferenceError', 'ResourceWarning', 'RuntimeError', 'RuntimeWarning', 'StopAsyncIteration', 'S
... (truncated - too long)
Full output: https://paste.pythondiscord.com/CHV2ZTOGRNC5CAD267FAMGO7OQ
azure flicker
And there's all the functions from all the modules in the stdlib.
storm coyote
no
stable creek
half of those are errors and warnings!
azure flicker
These are the "builtin names". The names you do not need to import.
stable creek
base R has like
azure flicker
This includes the predefined Exception types and all the builtin functions.
You can see min, max, print etc in there.
stable creek
mhm
I just find it strange that mathematics stuff are like
all in different packages?
Base R has let me check
heavy fern
base R has like
If you're coming from R, you'll find yourself in numpy a lot.
azure flicker
Ah, there's a few layers to that.
twin rivet
most of python is in its standard library (other 1st party packages)
stable creek
Im not coming from R, im learning both at the same time, but the R classes is moving a lot quicker than the python one
azure flicker
numpy is for performant bulk computation. Likewise pandas. There's a stdlib math module with basic math functions in it.
azure flicker
>>> import math
>>> dir(math)
['__doc__', '__file__', '__loader__', '__name__', '__package__', '__spec__', 'acos', 'acosh', 'asin', 'asinh', 'atan', 'atan2', 'atanh', 'ceil', 'comb', 'copysign', 'cos', 'cosh', 'degrees', 'dist', 'e', 'erf', 'erfc', 'exp', 'expm1', 'fabs', 'factorial', 'floor', 'fmod', 'frexp', 'fsum', 'gamma', 'gcd', 'hypot', 'inf', 'isclose', 'isfinite', 'isinf', 'isnan', 'isqrt', 'lcm', 'ldexp', 'lgamma', 'log', 'log10', 'log1p', 'log2', 'modf', 'nan', 'nextafter', 'perm', 'pi', 'pow', 'prod', 'radians', 'remainder', 'sin', 'sinh', 'sqrt', 'tan', 'tanh', 'tau', 'trunc', 'ulp']
this is the base R functions list
storm coyote
Base R has let me check
because R is specifically for mathematics / scientific stack
stable creek
ya
heavy fern
Im not coming from R, im learning both at the same time, but the R classes is mo...
It's more like how Python evolved. R evolved with a specialized purpose, Python evolved as a generalized thing that various specialties adopted.
stable creek
Python is considered kinda better overall
storm coyote
eh, depends.
azure flicker
It is probably more general purpose. I have not used R though.
stable creek
well hands down better overall if we're speaking in general terms
Like R is just stats and data
storm coyote
python is a general purpose language, R is specialized
stable creek
python is much more
azure flicker
Yeah. Off to numpy with you then!
stable creek
well it wont even be just numpy
for the things you could do in base R basically is like
Base R = numpy + pandas + scikit + matplotlib + seaborn + more
from what I can tell at least
twin rivet
well, consider using venv before installing more packages so you can isolate your versions
stable creek
and then R has further packages for more intensive data stuff?
heavy fern
Eh, you still need things like dplyr and other libs in R. It's not like it's one and done.
stable creek
for some stuff ya
idk if python does this
but R has like
some packages are a combination of other packages?
heavy fern
Well yah, many of those libraries all build on numpy
And many also build on other work, like cython
stable creek
if you install "Tidyverse" you get like 7 massive packages that cover everything that average data analyst or statistician would ever need
heavy fern
(Or numpy is also built on other stuff)
stable creek
Python seems hard with installing packages too :/
like all i have to do is tick a box in RStudio. No need for going into command prompt, importing constantly, etc
heavy fern
if you install "Tidyverse" you get like 7 massive packages that cover everything...
Not imo.
stable creek
Not imo.
well not everything
but its a huge portion
for data manipulation it covers everything, not really much for visualization though
can I get help making a common module? and help learning how to use it
heavy fern
Sorry, that was to the 'hard' point. I don't find it hard, or hard to explain. Most data scientists will work in notebook environments where we've already done most of the prep for them, and they can install libs via %pip magics
The tricky part is just learning to create a virtual environment (or even knowing that it's something you should do)
stable creek
i also dont know what that is
like I just use new files in Python IDLE
idk what virtual is, and idk how the scope of packages work
heavy fern
That's fine, you may switch to an IDE like Pycharm or VSCode. You may like notebooks better than idle.
stable creek
my class is using jupyter notebooks
but i HATE how its in a browser and not its own app
and I dont know any other IDEs
heavy fern
You can run the notebook from an app like the two I just mentioned
stable creek
whats a notebook
heavy fern
It's a sequence of 'code cells' that run in a Python interpreter
Think of it as one program broken into multiple blocks of code
stable creek
I actually have VScode installed
cause i was required to install anaconda
and it came with it
idk how to use it though
stable creek
uhhhh
heavy fern
Yup, file -> new
heavy fern
New file
stable creek
says select file type or name?
heavy fern
Yes
heavy fern
Yup, that is asking which Python version or environment (if you created any) you want to use
Because you might have multiple versions
stable creek
how do i check
heavy fern
Click select kernel
heavy fern
I'm not at my computer, I think it's first and third.
stable creek
hmm
uneven sirenBOT
Python help channel closed
