from pandas import HDFStore
from utils import dask_cov
from PyAI import *
with HDFStore("test.hdf5") as f:
Close = f.Close
close = Close.as_matrix()
open = f.Open.as_matrix()
names = f.Close.columns.tolist()
var = close - open
var -= var.mean(axis=0)
var /= var.std(axis=0)
var = var.T
# Create unsupervised labels
labels = cluster.AffinityPropagation(0.7).fit_predict(dask_cov(var))
brain = Brain(var, labels)
brain.init_data_transformation(TRANSFORMATION.Standardize(), TRANSFORMATION.PCA.RandomizedPCA())
brain.init_naive_bayes(NAIVE_BAYES.REGULAR)
from utils import dask_cov
# Gather data from databases
with h5py.File('close.hdf5') as f:
close_d = f['/quotes/close']
open_d = f['/quotes/open']
close = da.from_array(close_d, chunks=(100, 100))
open = da.from_array(open_d, chunks=(100, 100))
var = close - open
# Standardize data
var -= var.mean(axis=0)
var /= var.std(axis=0)
var = var.T
# Create unsupervised labels
labels = cluster.AffinityPropagation().fit_predict(dask_cov(var).compute())
store = HDFStore('quotes.hdf5')
names = store.quotes.minor_axis
brain = Brain(var.compute(), labels)
data = var.compute()
brain.init_data_transformation(TRANSFORMATION.Standardize(), TRANSFORMATION.PCA.RandomizedPCA())
brain.init_naive_bayes(NAIVE_BAYES.REGULAR)