def fit(self, data):
self.time = data.keys()
self.df = pd.Series(data)
self.df.index = pd.to_datetime(self.df.index, format='%Y-%m-%d')
self.df = pd.to_numeric(self.df, errors='coerce')
self.df = self.df.astype(float)
self.data_freq = infer_freq(self.df)
self.weight = self.compute_weight(self.df.index)
self.df.index.freq = self.data_freq
def fit(self, data, freq=None):
# self.time = data.keys()
self.df = pd.Series(data)
self.df.index = pd.to_datetime(self.df.index, format='%Y-%m-%d')
self.df = pd.to_numeric(self.df, errors='coerce')
self.df = self.df.astype(float)
if not freq:
freq = infer_freq(self.df)
self.data_freq = freq
self.df.index.freq = self.data_freq
def predict(self,time = 5):
self.freq = infer_freq(self.data)
print(self.freq)
future = self.model.make_future_dataframe(periods=time,freq =self.freq)
forecast = self.model.predict(future)
return forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']]
# data = {"2004-09-01": "14089", "2005-09-01": "5758", "2006-03-01": "3382", "2006-06-01": "3186", "2006-09-01": "3271", "2007-01-01": "5300", "2007-03-01": "20518", "2007-06-01": "18373", "2007-09-01": "53265", "2008-01-01": "91193", "2008-03-01": "223883", "2008-06-01": "197119", "2008-09-01": "237820", "2009-01-01": "208096", "2009-03-01": "100704", "2009-06-01": "92290", "2009-09-01": "60550", "2010-01-01": "44254", "2010-03-01": "110052", "2010-06-01": "157617", "2010-09-01": "117208", "2011-01-01": "92753", "2011-03-01": "64638", "2011-06-01": "154511", "2011-09-01": "151503", "2012-01-01": "136955", "2012-03-01": "176194", "2012-06-01": "196928", "2012-09-01": "194400", "2013-01-01": "240000", "2013-03-01": "296500", "2013-06-01": "306500", "2013-09-01": "385300", "2014-01-01": "402300", "2014-03-01": "309200", "2014-06-01": "387100", "2014-09-01": "330300", "2015-01-01": "438700", "2015-03-01": "415600", "2015-06-01": "541800", "2015-09-01": "731900", "2016-01-01": "1074900", "2016-03-01": "1010500", "2016-06-01": "1649000", "2016-09-01": "1309100", "2017-01-01": "810600", "2017-03-01": "986800", "2017-06-01": "2362600", "2017-09-01": "3003400", "2018-01-01": "5212800"}
# m = Prophet()
# m.fit(data)
# ans = m.predict(20)
# print(ans)
def transformArgsCheck(self, args):
try:
self.data = json.loads(args.data)
except Exception as e:
abort_if_transform_data_incorrect(e)
try:
self.od = OrderedDict(self.data)
except Exception as e:
abort_if_transform_data_incorrect(e)
startDate = list(islice(self.od.items(), 0, 1))[0][0]
endDate = list(islice(self.od.items(),
len(self.od) - 1, len(self.od)))[0][0]
self.targetStartDate = args.targetStartDate if args.targetStartDate else startDate
self.targetEndDate = args.targetEndDate if args.targetEndDate else endDate
if self.targetStartDate < startDate or self.targetEndDate > endDate:
abort_if_transform_time_incorrect()
self.type = args.type
self.targetFreq = args.targetFrequent
self.originFreq = args.originFrequent if args.originFrequent else infer_freq(
self.data)
def predict(self, time=5):
self.freq = infer_freq(self.data)
future = self.model.make_future_dataframe(periods=time, freq=self.freq)
forecast = self.model.predict(future)
self.time_dict = self._get_prediction_with_history(forecast)
return self.time_dict