#Stata #streamplot v1.6 is now out with several major changes:
- A cat() option added to compare streams across an additional binary (north vs south) variable category.
- Major code cleanup and rewrite of core routines to optimize speed.

https://github.com/asjadnaqvi/stata-streamplot

GitHub - asjadnaqvi/stata-streamplot: A Stata package for stream plots

A Stata package for stream plots. Contribute to asjadnaqvi/stata-streamplot development by creating an account on GitHub.

GitHub

Gonna up the #streamplot package a bit:
https://github.com/asjadnaqvi/stata-streamplot

Here you see a stream graph of top 50 female names in the USA from 1880-2014. Emma/Olivia/Sophia are the top 3, while Sarah/Hanna/Emily had their peaks. Elizabeth is high persistent!

More graphs below...

GitHub - asjadnaqvi/stata-streamplot: A Stata package for stream plots

A Stata package for stream plots. Contribute to asjadnaqvi/stata-streamplot development by creating an account on GitHub.

GitHub

The 
#streamplot v1.5 release is now out and has a major feature update:

Graphs can now be recentered in the middle (default), top, or bottom.

This allows the users a lot more flexibility in deciding how to visualize the information.

More info on GitHub:
https://github.com/asjadnaqvi/stata-streamplot

#stata #dataviz

GitHub - asjadnaqvi/stata-streamplot: A Stata package for stream plots

A Stata package for stream plots. Contribute to asjadnaqvi/stata-streamplot development by creating an account on GitHub.

GitHub

Version 1.4 of  #streamplot is now out:

https://github.com/asjadnaqvi/stata-streamplot.

Better color and label controls, better error checks, big improvements to the code in the backend.

#stata #dataviz

GitHub - asjadnaqvi/stata-streamplot: A Stata package for stream plots

A Stata package for stream plots. Contribute to asjadnaqvi/stata-streamplot development by creating an account on GitHub.

GitHub
#streamplot
変な経験則。
start_pointsで指定する点は+方向にずれて表示されるようだ。そのズレは分解能程度。従って現状の対処としては分解能の半分の値を引いた数値を与えるのが良さそうだ。例えば
x = np.linspace(-10, 10, 6)
y = np.linspace(-10, 10, 11)
の時start_pointsに(1,5)を設定したい場合は、(1-10/6,5-10/11)を指定する。ただしこの座標はstreamplotに与える時のみなので、グラフに点を打つ場合は(1,5)のままで。
結局のところこの引き算をしなくても分解能を高めれば相対的にズレは小さくなります。多分、内部的に座標からインデックスを計算する部分のバグですね。
#streamplot
ずれる問題。
例えばstart_pointsに(0,0)を指定した場合。
最初の
x = np.linspace(-10, 10, 10)
なんかだとxの列に0が含まれてなかったりだからじゃないかな?
#pyplot #streamplot
ちょいと実験してみました。
start_points で指定した複数の点が同じベクトル曲線上に乗る場合にエラーになるようです。