words-sdsc / edx Goto Github PK

I'm collecting the issues with git bash, then I'll add them as a note to the install instructions

• more and man are not available
• in Jupyter Notebooks, you cannot access git bash with !, so for example !ls won't work

someone suggested to use the Win 10 compatibility layer for Ubuntu, but I think that is too complicated.

@a1singh which one should I start from?

write instructions about required packages

@altintas do you have any week 7 material I can take a look at?

for meeting:

• what kind of notebook
• similar one I can use as an example
• which topics/packages should I focus on?

nbpresent?

intro to notebook

• @zonca - pandas
• @zonca - text data, 3 notebooks
• @a1singh - numpy
• @a1singh - Machine Learning, 3 notebooks, Classification, Clustering, Regression
• @a1singh - matplotlib week 5

Create notebooks on separate pull requests and merge them in.

@a1singh it looks like there is a visualization issue if you shrink the horizontal size of the notebook

@a1singh

• add details on how to download and extract data from https://www.kaggle.com/hugomathien/soccer

• Cell 5, better do df.shape in another cell, is not related to any row NULL?

• Maybe you can also take a look at missing data per column with df.isnull().sum(axis=0)?

• Cell 9 explain why you are shuffling

• Cell 14, write also feature name?

    for f in potentialFeatures:
         related = df['overall_rating'].corr(df[f])
          print(f + ":", related)
  

• First plot of correlations, kind of small labels, maybe just take a subset of those?

• At the end, in the clustering example, do you think you can describe in words what kind of groups are being identified?

• I am worried that parallel_plot is a bit difficult to explain and could distract from the most important points of the notebook, is there a way to achieve similar result in another way? maybe using seaborn?

@leporter when introducing Python executes some commands on the Python terminal, check if with the swcarpentry setup this works correctly on windows 7,8, 10.

• download from Kaggle is 370MB and quite slow, isn't it possible just to host the CSV somewhere? If not maybe just warn the students that this is going to take some time.
• you should not need to do tolist(), you can use len directly on the np.array
• I don't understand what you are doing in cell 34

In progress, I'll add more once I can run the notebook

Based on the content of the video scripts of Ilkay, prepare a reading that works on all platforms equally well.

@zonca : when you get time, please review the Pandas Notebook, thanks !

• This Notebook will use WIFIRE datasets & numpy

• small dataset
• viz
• machine learning
• Sloan?
• Hubble?
• Planck sources?

https://github.com/jupyter/docker-stacks/tree/master/scipy-notebook

Review Script-W3_L1_V5-UNIXCommandsInNotebooks

We might switch to Anaconda and Git Bash,

Check that it installs fine on:

• Windows 7
• Windows 8
• Windows 10

No need to test mac and linux.

Also test @altintas 's shell commands from GDrive in Git Bash

After install on Windows 10, at first launch I get this error. Troubleshooting:

Error response from daemon: A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond.

at Docker.Backend.DockerDaemonChecker.Check(Func`1 isDaemonProcessStillRunning)
at Docker.Backend.ContainerEngine.Linux.DoStart(Settings settings)
at Docker.Backend.ContainerEngine.Linux.Start(Settings settings)
at Docker.Core.Pipe.NamedPipeServer.<>c__DisplayClass8_0.b__0(Object[] parameters)
at Docker.Core.Pipe.NamedPipeServer.RunAction(String action, Object[] parameters)

crash reported:
A8167BAF-21AC-4D42-A015-54DA91463F6B/2017-04-09_12-30-58

@a1singh @leporter

we can create a header notebook with common cells to be prepended by nbgrader to the assigned notebooks:

http://nbgrader.readthedocs.io/en/stable/user_guide/creating_and_grading_assignments.html#assign-a-problem-set-using-nbgrader

should I go ahead and create it from the numpy notebook? or you prefer to take care of this?