DOC: Kernel density notebook update

[tommyod]

First PR to statsmodels. The Jupyter Notebook on Kernel Density Estimation seemed very incomplete to me, so I updated it with the following:

• More text and comment explaining the code.
• Example comparing a true distribution to the KDE of the sampled distribution.
• The effect of varying the bw argument (the bandwidth of the kernel).
• The various kernel functions available (one plot for each of them).
• Cleaned up the plots a bit: smaller, with grids, default maplotlib colors instead of black/red.

I also fixed a minor typo.

Comments very welcome!

[coveralls]

Coverage remained the same at 82.13% when pulling cd9b8cf on tommyod:kernel_density into 4151898 on statsmodels:master.

[coveralls]

Coverage remained the same at 82.13% when pulling cd9b8cf on tommyod:kernel_density into 4151898 on statsmodels:master.

[josef-pkt]

@tommyod Thanks for the improvement

I only had time for a very quick skimming of it.
Looks good overall
content of notebook will need to be removed before merging

the only comment directly to the content is that icdf doesn't fit in the plot with cdf and sf.
The x-axis for icdf is misleading because it should be the interval [0, 1]

[codecov-io]

Codecov Report

❗ No coverage uploaded for pull request base (master@91ed779). Click here to learn what that means.
The diff coverage is n/a.

@@            Coverage Diff            @@
##             master    #4207   +/-   ##
=========================================
  Coverage          ?   79.55%           
=========================================
  Files             ?      562           
  Lines             ?    83816           
  Branches          ?     9551           
=========================================
  Hits              ?    66683           
  Misses            ?    14947           
  Partials          ?     2186

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 91ed779...b357286. Read the comment docs.

[coveralls]

Changes Unknown when pulling d475bf4 on tommyod:kernel_density into ** on statsmodels:master**.

[coveralls]

Changes Unknown when pulling b357286 on tommyod:kernel_density into ** on statsmodels:master**.

[coveralls]

Changes Unknown when pulling b357286 on tommyod:kernel_density into ** on statsmodels:master**.

[tommyod]

@josef-pkt Thanks for the review. I read through again and wrote a little bit more. I removed the output of the Jupyter Notebook, and addressed the comment about the domain of the inverse CDF as per your request :)

[josef-pkt]

I think using alpha=0.5 for the histograms except the first would look better, better visibility of plot lines on top of histogram

[josef-pkt]

typo mixture

[josef-pkt]

A squashed and rebased version of this is now in #4551

[josef-pkt]

merged in #4551

@tommyod Thanks for the improvement