search

NYUDataBootcamp / Book

Textbook to accompany class
Creative Commons Attribution 4.0 International
7 stars 4 forks source link

Question: should we introduce pandas_datareader in pandas-input.md? #13

Closed sglyon closed 7 years ago

sglyon commented 8 years ago

In this section of the pandas-input.md chapter we go over how to use the API methods from pandas.io to read in data from FRED, IMF, and others.

That code has been deprecated in pandas for some time and users are told to use the pandas_datareader package.

Eventually the deprecated code will be deleted altogether from pandas and we'll be forced to use pandas_datareader to read in the data.

I think the obvious answer is that we should use the currently suggested method here instead of the deprecated routines, but this brings up the issue of using conda to install a package that doesn't come with Anaconda -- something we haven't talked about at this point in the book.

What do people think? Should we take the time to talk about pip and conda before this section (maybe at the start of this lecture in our discussion on packages) so we can use pandas_datareader?

cc7768 commented 8 years ago

Definitely agree. I think goal is to keep material up to date. It would be unfortunate for students to learn things that will go away in future.

We can introduce conda and pip when we do installation? Just talk about its role as a package manager etc...

cc7768 commented 7 years ago

Related to this:

We need to go through and change all of the import pandas.io.data as web to the new correct imports from pandas_datareader.

cc7768 commented 7 years ago

Addressed by https://github.com/NYUDataBootcamp/Book/pull/22#pullrequestreview-5888600