Regression of FOPDT Parameters in Python

[@Kakikiwi-eu2kr]

You make a separate call odeint for each timestep. Using one single big call to odeint (or solve_ivp now), specifying the required time evaluations, does not work as well. It goes much faster, but the resulting optimization is not as good. The simulated curves also look a bit like there's some Runge effect. Why is it so much better to integrate point by point ?

[@apm]

Sometimes the error control doesn’t work well and smaller forced steps help. Thanks for the observation.

[@rrc]

Instead of using wget, you can how just import the file with Pandas directly from the URL. See apmonitor.com/pdc/index.php/Main/DynamicParameterRegression?action=sourceblock&num=1 for an example of how to do this.

[@pkl520]

hope u can turn on youtube's subtitles on this vid. big tks~!

[@apm]

The auto-generated subtitles are available now.

[@pkl520]

big thanks~~~!