Gotcha - use pandas_profiling see some pit EDA data document encountered

A recent study characteristic project -related stuff and found pandas_profilingthis package is particularly suitable for visual data to be able to show where many of the details, including the missing values and other relevant information. Then, I started tossing pip install pandas_profilingdiscovered all of a sudden had good environment ( Anaconda3.5, A Python 3 ) to toss the bad ( above expression package represents the time my mood ---- thin blue mushrooms ~ ), and the emergence of the following a few questions, which I made records, and share out today -

Question one

ImportError: No module named 'win32api'。

For this first issue, the main problem is more common in the windows system, the first reaction is to not install the package, in order to determine their own conjecture, using pip liststart a look and found the environment does exist in this package, but it is not import. Then with a full doubt, start the search in the browser, try various solutions and finally do a little summary -

First, it may be that there is no package or download the full compatibility issues, then you can try to uninstall and then reinstall;
Second, if the re-installation pip install pywin32process of a problem could not find a version that statists the requirment win32api..., this time to replace download method pip install pypiwin32can be.

Question two

Jupyter notebook kernel not connecting。

ValueError: signal only works in main thread
ERROR:tornado.application:Exception in callback <functools.partial object at 0x111e2b208>
Traceback (most recent call last):

This problem occurs after a problem, the most likely cause is a process of finding solutions to problems to do an pip update...operation, leading Tornadoand Jupyterversions or pyzmqand ipykernelincompatibility phenomenon.

For me the problem is mainly due to pyzmqand ipykernelincompatible cause, the solution:

pip install "pyzmq==17.0.0" "ipykernel==4.8.2"

Question three

Jupyter notebook “IOPub data rate exceeded.”

     IOPub data rate exceeded.
The notebook server will temporarily stop sending output
to the client in order to avoid crashing it.
To change this limit, set the config variable
`--NotebookApp.iopub_data_rate_limit`.

The problem is input ppf.ProfileReport(train_data)caused mainly due to the pandas_profilingcontents of the file packet output is too large, more than the default limit in order to prevent system crashes, computer servers stopped this action accordingly.

solution:

1. generation jupyter_notebook_config.py, will generate the corresponding file in the corresponding directory;

To create a jupyter_notebook_config.py file, with all the defaults commented out, 
you can use the following command line:

$ jupyter notebook --generate-config

2. Open the file and search c.NotebookApp.iopub_data_rate_limit;
3. Comment out the line c.NotebookApp.iopub_data_rate_limit = 1000000and change it to a higher default rate. (I changed to c.NotebookApp.iopub_data_rate_limit = 2147483647).

Hint: This problem Jupyter notebook release 5.2.2version done to solve, can be appropriately selected update! ! !

PS: the problem-solving process, feel like at work, constantly tossing ,, problem solving, and I feel it was a Bel cool Ha! ! To sum it up: No Zuo No Died!