def first_principal_component_SGD(data):
""" Uses SGD method to determine the direction that maximizes the
directional_variance gradient """
guess = data[0]
unscaled_maximizer, _ = Ch8.maximize_stochastic(lambda x,_,w:
directional_variance_i(x,w), lambda x,_,w:
directional_variance_gradient_i(x,w), data,
[None for _ in data], guess)
return direction(unscaled_maximizer)
def first_principal_component_SGD(data):
""" Uses SGD method to determine the direction that maximizes the
directional_variance gradient """
guess = data[0]
unscaled_maximizer, _ = Ch8.maximize_stochastic(
lambda x, _, w: directional_variance_i(x, w),
lambda x, _, w: directional_variance_gradient_i(x, w), data,
[None for _ in data], guess)
return direction(unscaled_maximizer)
def first_principal_component(data):
guess = [1 for _ in data[0]]
# use partial to make the target and grade fncs a variable of w only
unscaled_maximizer, _, _ = Ch8.maximize_batch(
partial(directional_variance, data),
partial(directional_variance_gradient, data),
guess,tolerance = 0.00001)
return direction(unscaled_maximizer)
def first_principal_component(data):
guess = [1 for _ in data[0]]
# use partial to make the target and grade fncs a variable of w only
unscaled_maximizer, _, _ = Ch8.maximize_batch(
partial(directional_variance, data),
partial(directional_variance_gradient, data),
guess,
tolerance=0.00001)
return direction(unscaled_maximizer)