def gradient(X, Y, gd_lambda, descent=True, max_iterations=1000, binary=False):
accepted = False
iterations = 0
if type(X) is pd.DataFrame:
X = pandas_to_data(X)
# add column of ones
X = add_col(X, 1)
w = [0 for _ in range(len(X[0]))]
hrows = [0 for _ in range(len(X))]
print type(list(Y))
not_converged = len(X)
while iterations < max_iterations and not_converged > 0:
not_converged = len(X)
iterations += 1
for r in range(len(X)): # r = i, c = j
row = X[r]
h = get_hw(row, w, binary)
hrows[r] = h
#TODO this doesn't seem right!
#print 'values: {} {} '.format(list(Y)[r], h)
if h-list(Y)[r] == 0:
not_converged -= 1
for c in range(len(row)):
w[c] = w[c] - (gd_lambda * (h - list(Y)[r]) * row[c])
debug_print(iterations, not_converged, hrows, Y)
return w
def test_add_col():
array = []
for i in range(5):
row = []
for j in range(3):
row.append(0)
array.append(row)
print array
array = utils.add_col(array, 1)
print array