Content deleted Content added
No edit summary |
m bug in python algorithm: we need to save the "better model", which is trained on the maximum number of datapoints, not the "maybe model", which is just the initial guess based on the minimal number of points. |
||
Line 125:
if this_error < self.best_error:
self.best_error = this_error
self.best_fit =
return self
| |||