search

TheodoreEhrenborg / mercantilism

MIT License
0 stars 0 forks source link

I need a good way to estimate confidence. #21

Closed TheodoreEhrenborg closed 5 years ago

TheodoreEhrenborg commented 5 years ago

Use Decimal and write my own integrate(). Be clever.

TheodoreEhrenborg commented 5 years ago

For faster programming, I'll write everything in Python first. So the present contents of bayesian.py are going to change.

TheodoreEhrenborg commented 5 years ago

That was just done in Commit 7653bc97a9009e936901cbf330ced157066390b8

TheodoreEhrenborg commented 5 years ago

With cdecimal, it got about 10 times faster. Most of the time is spent in the both() method. I'm not sure if that time is spend inside c_decimal, but the least I can do is speed up both() by adding more types. Almost all of the rest of the time is spend in integrate(). Cython is at least 10% faster than Python. I also need to add a connection from API to c_bayesian.

TheodoreEhrenborg commented 5 years ago

See Issue #8 for earlier thoughts.

TheodoreEhrenborg commented 5 years ago

I finished this two weekends ago. The confidence levels tend to vary by about 10 decibels, but with the square-root method I can create solid confidence measurements.