DOC: astype now takes dict mapping col names to datatypes (#14761)

[m-charlton]

• closes DOC: update docs for .astype with dict #14761
• tests added / passed
• passes git diff upstream/master | flake8 --diff
• whatsnew entry

Updating documentation to reflect change

[jorisvandenbossche]

@m-charlton Thanks for working on this!
I added a few comments (which I now see you already asked about this in the issue, but only see it now)

[jorisvandenbossche]

0.20 -> 0.19

and I would also clarify what was added in 0.20.0 (only the ability to specify different dtypes for different columns with a dict)

[m-charlton]

I think I understand this

[m-charlton]

I'm a little confused. I tried using a dict with column/dtype mapping as an
argument to astype and this works in 0.19.0:

>>> import numpy as np
>>> import pandas as pd
>>> pd.__version__
'0.19.0'
>>> dft = pd.DataFrame({'a': [1,0,1], 'b': [4,5,6], 'c': [7, 8, 9]})
>>> dft = dft.astype({'a': np.bool, 'c': np.uint8})
>>> dft
       a  b  c
0   True  4  7
1  False  5  8
2   True  6  9
>>> dft.dtypes
a     bool
b    int64
c    uint8
dtype: object

I think I'm missing what new functionality has been introduced in v0.20

[jorisvandenbossche]

@m-charlton As I said above, this feature was already introduced in 0.19.0. That's the reason I asked to change 0.20.0 to 0.19.0 in the versionadded here

[m-charlton]

That makes sense. Will submit changes today.

[jorisvandenbossche]

Can you leave the previous example there as well, and add this line (the single dtype is still a valid case)

[jorisvandenbossche]

So I would leave this example, but just add a new one below this one, also beginning with an explaining sentence, like "Convert certain columns to as specific dtype by passing a dict to astype: ..."

[m-charlton]

Will make relevant changes and resubmit. Old example will be left in place as is.

[jorisvandenbossche]

@m-charlton This is not needed, as this enhancement was already added in 0.19.0, and is mentioned in the whatsnew of 0.19.0 (see the changes in the PR https://github.com/pandas-dev/pandas/pull/13375/files)

[m-charlton]

Will leave this file unchanged.

[codecov-io]

Current coverage is 84.64% (diff: 100%)

Merging #14837 into master will decrease coverage by <.01%

@@             master     #14837   diff @@
==========================================
  Files           144        144          
  Lines         51016      51016          
  Methods           0          0          
  Messages          0          0          
  Branches          0          0          
==========================================
- Hits          43185      43184     -1   
- Misses         7831       7832     +1   
  Partials          0          0          

Powered by Codecov. Last update 8e630b6...b9414a9

[jorisvandenbossche]

@m-charlton you don't need to close this PR or remove the branch on github to update this. You can just push to the same branch, and the PR on github will automatically be updated.
If you restore the branch, reopen the PR, and then push again, it should still work like that.

[jorisvandenbossche]

@m-charlton Some git help, seems something went wrong with updating to master. If you do the following, it should normally be solved:

git fetch upstream
git checkout update_docs_astype_with_dict_#14761
git rebase upstream/master
git push -f 

[m-charlton]

I performed the above before I saw your git help.
Has this fixed the problem?

[m-charlton]

I think I've rectified my mistakes in update b9414a9

[jreback]

lgtm. ping on green.

[m-charlton]

I think it was more to do with the content so @jorisvandenbossche needs to
give it a look.