CosmoPhotoz 0.1

Features

• Principle Component Anylsis and decomposition of input photometric catalogue

• Generalized Linear Model family and link choice

• Seaborn publication quality plots

Get it now

The package can be installed using the PyPI and pip.

$ pip install -U CosmoPhotoz

Or if the tarball or repository is downloaded, distutils can be

$ python setup.py install

Examples

Run from the command line.

$ run_glm.py --dataset sample.csv --num_components 3 --training_size 10000 --family Gamma --link log

Or import the library into python.

from CosmoPhotoz.photoz import PhotoSample # import the library
import numpy as np

# Instantiate the class
UserCatalogue = PhotoSample(filename="PHAT0", family="Gamma", link="log")

# Make a training size array to loop through
train_size_arr = np.arange(500,10000,500)
catastrophic_error = []

# Select your number of components
UserCatalogue.num_components = 4

for i in range(len(train_size_arr)):
    UserCatalogue.do_PCA()
    UserCatalogue.test_size = train_size_arr[i]
    UserCatalogue.split_sample(random=True)
    UserCatalogue.do_GLM()
    catastrophic_error.append(UserCatalogue.catastrophic_error)

min_indx = np.array(catastrophic_error) < 5.937
optimum_train_size = train_size_arr[min_indx]
print optimum_train_size

See more examples within the Documentation.

Documentation

• The library documentation can be accessed at Read the Docs

• The git repository can be accessed at GitHub

• The PyPI package page can be accessed at PyPI

Requirements

• Python >= 2.7 or >= 3.3

License

• GNU General Public License (GPL>=3)