https://github.com/esheldon/bootstrap

python code to do bootstraps, wrapping fast C code
https://github.com/esheldon/bootstrap

Last synced: 2 days ago
JSON representation

python code to do bootstraps, wrapping fast C code

• Host: GitHub
• URL: https://github.com/esheldon/bootstrap
• Owner: esheldon
• Created: 2015-09-26T13:21:36.000Z (about 9 years ago)
• Default Branch: master
• Last Pushed: 2015-09-26T20:38:00.000Z (about 9 years ago)
• Last Synced: 2024-10-19T19:43:35.279Z (19 days ago)
• Language: C
• Size: 156 KB
• Stars: 0
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# bootstrap

python code to do bootstraps, wrapping fast C code

Working in C we can avoid large memory allocations for the bootstrap subsets

```python

from bootstrap import bootstrap, Bootstrap

mean = array(  [2.5, -5.6] )

cov  = array([ [1.5,  0.2],

               [0.4,  2.7] ])

# create correlated data using cholesky sampling

np = 10000

rdata = cholesky_sample(mean,cov, np)

nboot=1000

ecov = numpy.cov(rdata.T)

print('expected mean:',mean)

print('expected cov: ')

print(ecov/np)

res=bootstrap(rdata, nboot)

print('boot mean:',res['mean'])

print('boot cov:')

print(res['cov'])

# the result is a dict with the following entries

#    A dictionary with entries

#            'mean': mean over the sample

#            'cov': covariance over the sample

#            'err': sqrt of the diagonal of the covariance matrix

#            'seed': the used seed

#            'nboot': number of bootstrap realizations

 

# You can also use an object

b=Bootstrap(data)

b.go(nboot)

res = b.get_result()

```