Package to conduct factor analysis on data
Project description
Factor Analysis
Your data is factorized into latent variables and noise parameters all within the same sample.
m
denotes sample length,
n
denotes number of features for the data sample
k
denotes number of latent features to be represented for the data sample
λ
denotes the factor
z
denotes the latent variable of size m
x k
ϵ
denotes the noise parameters of size m
x ``n
ψ` denotes the covariance of `ϵ`
Factor Analysis equation
x = μ + λz + ϵ
We determine λ
and ψ
using posterior distribution ( z | x ) by expectation maximisation. The method is useful to predict the factor variables from a posterior distribution known to the user provided the data you are processing can be fit into the equation.
import tensorflow as tf
f = factor_analysis.factors.Factor(data, factor_analysis.posterior.Posterior(covariance_prior, means))
noise = factor_analysis.noise.Noise(f, f.posterior)
with tf.Session() as sess:
print(f.create_factor().eval())
print(noise.create_noise(f.create_factor()).eval())
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for factor_analysis-0.0.2-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 9f490e73095e9e4c34d10eea3d8fb135cb97cbd9a5276d4267045296890e82df |
|
MD5 | 068783dae81232471c21e1871199ce47 |
|
BLAKE2b-256 | b97f8f0a34ea30b4457514d54d69e5a69ea3b9d5ab64f4ba82eaebfd678cb461 |