def p2SLR_predict(tData_xy, base_date, end_date, predict_start_date,
predict_end_date, flavor_names):
# Make prediction
predict_start_x = parser.cal_day(base_date, predict_start_date)
predict_end_x = parser.cal_day(base_date, predict_end_date)
predict_flavors = []
tDays = parser.cal_day(base_date, end_date)
pDays = parser.cal_day(base_date, predict_end_date) - parser.cal_day(
base_date, predict_start_date)
flavors_wb = {}
new_tData_xy = {}
for key in tData_xy.keys():
new_tData_xy[key] = tData_xy[key][-7:]
flavors_wb = get_flavors_wb(new_tData_xy)
print 'flavors_wb:', flavors_wb
for each in flavor_names:
num = tData_xy[each][-1][1]
# get the num of the following weekend's machines ####
pflavor_num = (tDays +
pDays) * flavors_wb[each][0] + flavors_wb[each][1] - num
pflavor_num *= 1.1
if pflavor_num < 0:
pflavor_num = 0
elif pflavor_num - int(pflavor_num) < 0.5:
pflavor_num = int(pflavor_num) + 1
else:
pflavor_num = int(pflavor_num)
print ' pflavor_num: ', pflavor_num
predict_flavors.append((each, pflavor_num))
return predict_flavors
def ls_predict(tData_xy=None,
base_date=None,
end_date=None,
predict_start_date=None,
predict_end_date=None,
flavor_names=None):
predict_flavors = []
tDays = parser.cal_day(base_date, end_date)
pDays = parser.cal_day(base_date, predict_end_date) - parser.cal_day(
base_date, predict_start_date)
# print tData_xy
new_tData_xy = {}
for each in flavor_names:
x = []
y = []
new_tData_xy[each] = tData_xy[each][-pDays:]
for t in new_tData_xy[each]:
x.append(t[0])
y.append(t[1])
a, b = calcAB(x, y)
print a, b
py = a * (tDays + pDays) + b
if py < tData_xy[each][-1][1]:
num = tData_xy[each][-1][1]
num = int(round(num * pDays / tDays))
else:
num = py - tData_xy[each][-1][1]
#num = int(round(num*pDays/tDays))
predict_flavors.append((each, int(num)))
print predict_flavors
return predict_flavors
def average_predict(tData_xy=None,
base_date=None,
end_date=None,
predict_start_date=None,
predict_end_date=None,
flavor_names=None):
predict_flavors = []
tDays = parser.cal_day(base_date, end_date)
pDays = parser.cal_day(base_date, predict_end_date) - parser.cal_day(
base_date, predict_start_date)
for each in flavor_names:
num = tData_xy[each][-1][1]
num = num * pDays / tDays
predict_flavors.append((each, num))
return predict_flavors
def p1SLR_predict(tData_xy, base_date, end_date, predict_start_date,
predict_end_date, flavor_names):
# Make prediction
predict_start_x = parser.cal_day(base_date, predict_start_date)
predict_end_x = parser.cal_day(base_date, predict_end_date)
predict_flavors = []
tDays = parser.cal_day(base_date, end_date)
pDays = parser.cal_day(base_date, predict_end_date) - parser.cal_day(
base_date, predict_start_date)
flavors_wb1 = {}
flavors_wb2 = {}
new_tData_xy1 = {}
new_tData_xy2 = {}
for key in tData_xy.keys():
new_tData_xy1[key] = tData_xy[key][-7:]
new_tData_xy2[key] = tData_xy[key][-4:]
print tData_xy[key]
print new_tData_xy1[key]
print new_tData_xy2[key]
print '-' * 20
flavors_wb1 = get_flavors_wb(new_tData_xy1)
flavors_wb2 = get_flavors_wb(new_tData_xy2)
for each in flavor_names:
num = tData_xy[each][-1][1]
# get the num of the following weekend's machines ####
pflavor_num1 = (
tDays + pDays) * flavors_wb1[each][0] + flavors_wb1[each][1] - num
pflavor_num2 = (
tDays + pDays) * flavors_wb2[each][0] + flavors_wb2[each][1] - num
rate = 0.598
pflavor_num = pflavor_num1 * rate + pflavor_num2 * (
1 - rate) + random.randint(-1, 1)
if pflavor_num < 0:
pflavor_num = 0
elif pflavor_num - int(pflavor_num) < 0.5:
pflavor_num = int(pflavor_num) + 1
else:
pflavor_num = int(pflavor_num)
print ' pflavor_num: ', pflavor_num
predict_flavors.append((each, pflavor_num))
return predict_flavors
def SLR_predict(tData_xy, base_date, end_date, predict_start_date,
predict_end_date, flavor_names):
#for key in tData_xy.keys():
#new_tData_xy[key] = tData_xy[key][-14:-1]
#print 'new_data_xy:', new_tData_xy
#flavors_wb = get_flavors_wb(new_tData_xy)
# Make prediction
predict_start_x = parser.cal_day(base_date, predict_start_date)
predict_end_x = parser.cal_day(base_date, predict_end_date)
predict_flavors = []
tDays = parser.cal_day(base_date, end_date)
pDays = parser.cal_day(base_date, predict_end_date) - parser.cal_day(
base_date, predict_start_date)
#get the everyday of the predicted day's flavors number
for day in range(pDays):
for each in flavor_names:
new_tData_xy = {}
for key in tData_xy.keys():
new_tData_xy[key] = tData_xy[key][-7 - day:]
flavors_wb = get_flavors_wb(new_tData_xy)
wangNum = (tDays + day +
1) * flavors_wb[each][0] + flavors_wb[each][1]
if wangNum < 0:
wangNum = 0
elif wangNum - int(wangNum) < 0.5:
wangNum = int(wangNum) + 1
else:
wangNum = int(wangNum)
new_tData_xy[each].append((tDays + day + 1, wangNum))
#print 'new ....' ,new_tData_xy
# get the num of the following weekend's machines ####
for each in flavor_names:
#print 'new_tData_xy[', each , '][-1][1]', new_tData_xy[each][-1][1]
#print 'new_tData_xy[each][-8][1]', new_tData_xy[each][-8][1]
pWeek_flavor_num = new_tData_xy[each][-1][1] - new_tData_xy[each][-8][1]
predict_flavors.append((each, pWeek_flavor_num))
#print 'predict_flavors:', predict_flavors
return predict_flavors
def tls_predict(tData_xy=None,
base_date=None,
end_date=None,
predict_start_date=None,
predict_end_date=None,
flavor_names=None):
predict_flavors = []
tDays = parser.cal_day(base_date, end_date)
pDays = parser.cal_day(base_date, predict_end_date) - parser.cal_day(
base_date, predict_start_date)
# print tData_xy
new_tData_xy = {}
for each in flavor_names:
x = []
y = []
new_tData_xy[each] = tData_xy[each][-7:]
for t in new_tData_xy[each]:
x.append(t[0])
y.append(t[1])
a, b, y_begin, y_end = calcAB2(x, y)
print a, b
mid = (x[0] + x[-1]) / 2
half = mid - x[0]
emm = (tDays + pDays - mid) / half
py = a * (emm) * 1.0273 + b
py = restore_scale1([py], y_begin, y_end)[0]
if py < tData_xy[each][-1][1]:
num = tData_xy[each][-1][1]
num = int(round(num * pDays / tDays))
else:
num = py - tData_xy[each][-1][1]
#num = int(round(num*pDays/tDays))
predict_flavors.append((each, int(num)))
print predict_flavors
return predict_flavors
def average_predict(tData_xy=None,
base_date=None,
end_date=None,
predict_start_date=None,
predict_end_date=None,
flavor_names=None):
predict_flavors = []
tDays = parser.cal_day(base_date, end_date)
pDays = parser.cal_day(base_date, predict_end_date) - parser.cal_day(
base_date, predict_start_date)
for each in flavor_names:
num = tData_xy[each][-1][1] - tData_xy[each][1][1]
d = tData_xy[each][-1][0] - tData_xy[each][-4][0]
if d == 0:
num = int(num * pDays / tDays)
else:
num = int(num * pDays / d)
predict_flavors.append((each, num))
return predict_flavors