⚠ WARNING: This package is very much a work in progress
pyby is a re-write of the {BFDA} R
Package.
In the future this library will be extended into a full package, but the current version is just intended as a demonstration of how much compute time can be reduced through the use of better algorithms and a switch in language.
On my blog I outlined a functional re-write of the BFDA package. In Part I and Part II I demonstrated how adopting a more functional coding style had a duel benefit: First, it resulted in more readable code. And Second, it made it easier to identify computational inefficiencies.
Using these principles I re-wrote the {BFDA} package in R. This led
to a large reduction in the time needed to perform the simulations (see
table below).
| VERSION | RUNNING TIME |
|---|---|
| BFDA (original) | 397.119s (~6.62 minutes) |
| tidyverse-based rewrite | 95.661s (~1.59 minutes) |
| base R-based rewrite | 77.866s (~1.3 minutes) |
This package demonstrates a re-write in Rust with an even large decrease in compute time. For same model as outlined in the blog post in the Python/Rust version completes the simulations in 1.6s. The results are shown below.
import pyby
import time
import pandas as pd
tic = time.perf_counter()
df = pyby.bf_sim(
0.5,
"t.paired",
{"family": "Cauchy", "params": [0, 0.707], "alternative": "two.sided"},
sampling_rule={"n_min": 50, "n_max": 300, "step_size": 5},
alternative="two.sided",
reps=1000,
seed=1600,
)
df = pd.DataFrame.from_records(
df, columns=["id", "true.ES", "n", "logBF", "emp.ES", "statistic", "p.value"]
)
toc = time.perf_counter()
print(f"Simulation performed in {toc - tic:0.4f} seconds")
print(df)Simulation performed in 1.6360 seconds
| | id | true.ES | n | logBF | emp.ES | statistic | p.value |
|-------|------|--------- |------|-----------|----------|-----------|--------------|
| 0 | 0 | 0.5 | 50 | 5.043586 | 0.580265 | 4.103096 | 1.535430e-04 |
| 1 | 0 | 0.5 | 55 | 4.275476 | 0.515145 | 3.820419 | 3.461539e-04 |
| 2 | 0 | 0.5 | 60 | 5.397193 | 0.539404 | 4.178204 | 9.844347e-05 |
| 3 | 0 | 0.5 | 65 | 3.790283 | 0.450784 | 3.634336 | 5.565287e-04 |
| 4 | 0 | 0.5 | 70 | 3.393479 | 0.417477 | 3.492866 | 8.385545e-04 |
| ... |... | ... | ... | ... | ... | ... | ... |
| 50995 | 999 | 0.5 | 280 | 29.076271 | 0.507505 | 8.492179 | 1.224343e-15 |
| 50996 | 999 | 0.5 | 285 | 28.385162 | 0.496423 | 8.380589 | 2.467588e-15 |
| 50997 | 999 | 0.5 | 290 | 27.564108 | 0.484431 | 8.249554 | 5.680903e-15 |
| 50998 | 999 | 0.5 | 295 | 26.432018 | 0.469911 | 8.070985 | 1.798275e-14 |
| 50999 | 999 | 0.5 | 300 | 26.020612 | 0.461992 | 8.001939 | 2.724417e-14 |
[51000 rows x 7 columns]
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent