def orgPrice_data(params):
columns = ["orgPrice_id", "fc", "orgPrice"]
with open(params["localFileName_org_orgPrice_data"], 'w') as f:
f.write(','.join(columns))
f.write('\n')
for sample in cursor1.find(
{'date': {
'$gte': datetime.today().strftime("%Y-%m-%d")
}}):
orgPrice_id = sample.get('_id')
del sample['_id']
del sample["date"]
del sample["ut"]
del sample["src"]
for key, value in sample.items():
try:
orgPrice = value.get('Y').get('price')
content = ','.join([orgPrice_id, key, orgPrice])
f.write(content + '\n')
except:
continue
logger.info("====\"{}\" finished====".format(
params["localFileName_org_orgPrice_data"].split('/')[-1]))
utils.delete_before4_localData(params["localFileName_org_orgPrice_data"],
params)
utils.upload_to_hdfs(params["localFileName_org_orgPrice_data"],
params["sparkDirName_org_orgPrice_data"], params)
def infoBase_data(params):
columns = ["infoBase_id", "departTime", "arriveTime", "isShare"]
with open(params["localFileName_org_infoBase_data"], 'w') as f:
f.write(','.join(columns))
f.write('\n')
for sample in cursor4.find(
{'date': {
'$gte': datetime.today().strftime("%Y-%m-%d")
}}):
infoBase_id = sample.get("_id")
departtime = sample.get('origindeparttime')
arrivetime = sample.get('originarrivetime')
isShare = sample.get('isshare')
try:
content = ','.join(
[infoBase_id, departtime, arrivetime,
str(isShare)])
f.write(content + '\n')
except:
continue
logger.info("====\"{}\" finished====".format(
params["localFileName_org_infoBase_data"].split('/')[-1]))
utils.delete_before4_localData(params["localFileName_org_infoBase_data"],
params)
utils.upload_to_hdfs(params["localFileName_org_infoBase_data"],
params["sparkDirName_org_infoBase_data"], params)
def lowPrice_data_main(params):
write_idList(params, cursor)
utils.delete_before4_localData(params["localFileName_lowPrice_idList"],
params)
if os.path.exists(params["localFileName_org_lowPrice_data"]):
os.remove(params["localFileName_org_lowPrice_data"])
columns = ["queryDate", 'price', 'id', 'org', 'dst']
with open(params["localFileName_org_lowPrice_data"], 'a') as f_write:
f_write.write(",".join(columns))
f_write.write("\n")
f_write.seek(2)
p = Pool(10)
counter = 0
with open(params["localFileName_lowPrice_idList"], 'r') as f_read:
for line in f_read:
counter += 1
L = line.strip().split(',')
p.apply_async(lowPrice_data, args=(
params,
L,
))
p.close()
p.join()
logger.info("=====\"{}\" finished======".format(
params["localFileName_org_lowPrice_data"].split('/')[-1]))
utils.delete_before4_localData(params["localFileName_org_lowPrice_data"],
params)
utils.upload_to_hdfs(params["localFileName_org_lowPrice_data"],
params["sparkDirName_org_lowPrice_data"], params)
def seatleft_data(params):
columns = ['queryDatetime', 'seatLeft', 'seatLeft_id']
with open(params["localFileName_org_seatLeft_data"], 'w') as f:
f.write(','.join(columns))
f.write('\n')
f.seek(2)
for sample in cursor3.find({}):
seatLeft_id = sample.get('_id')
df = pd.DataFrame.from_dict(
sample.get('fc'),
orient='index').reset_index().rename(columns={
'index': 'queryDatetime',
0: 'seatLeft'
})
df['seatLeft_id'] = seatLeft_id
df.to_csv(params["localFileName_org_seatLeft_data"],
header=False,
index=False,
mode='a')
logger.info("====\"{}\" finished====".format(
params["localFileName_org_seatLeft_data"].split('/')[-1]))
utils.delete_before4_localData(params["localFileName_org_seatLeft_data"],
params)
utils.upload_to_hdfs(params["localFileName_org_seatLeft_data"],
params["sparkDirName_org_seatLeft_data"], params)
def lowPrice_online_data(params):
tomorrow_str2 = datetime.strftime(
params["generateDate"] + timedelta(days=1), '%Y-%m-%d')
after30_str2 = datetime.strftime(
params["generateDate"] + timedelta(days=30), '%Y-%m-%d')
monthDay_list = pd.date_range(
start=tomorrow_str2, end=after30_str2,
freq='d').map(lambda x: datetime.strftime(x, '%m-%d'))
columns = ["queryDate", 'price', 'id', 'org', 'dst']
with open(params["localFileName_org_lowPrice_onlineData"], 'w') as f:
f.write(','.join(columns))
f.write('\n')
for monthDay in monthDay_list:
#对于online数据，根据id的条件筛选，然后只要拿到最大查询日期对应的价格即可（一条记录）。进一步的判断信息在spark上完成
#如果最大查询日期距离今天相隔7天以内，即该id在最近7天有价格记录，则将最大查询日期改为yesterday（模型只能通过昨天的信息，预测今天之后的价格趋势）
for sample in cursor.find(
{'_id': {
"$regex": r'.*{}$'.format(monthDay)
}}):
lowPrice_id = sample.get('_id')
del sample['_id']
org = sample.get('dairport')
del sample['dairport']
dst = sample.get('aairport')
del sample['aairport']
# df = pd.DataFrame.from_dict(sample, orient='index').reset_index().rename(
# columns={'index': 'queryDate', 0: 'price'})
queryDate = max(sample.keys())
price = sample.get(queryDate)
# historyLowPrice_fn_domestic中有异常，例如_id="3U3100_null_null_09-24"
try:
content = ','.join(
[queryDate, price, lowPrice_id, org, dst])
f.write(content + '\n')
except:
continue
logger.info("=====\"{}\" finished======".format(
params["localFileName_org_lowPrice_onlineData"].split('/')[-1]))
utils.delete_before4_localData(
params["localFileName_org_lowPrice_onlineData"], params)
utils.upload_to_hdfs(params["localFileName_org_lowPrice_onlineData"],
params["sparkDirName_org_lowPrice_onlineData"],
params)
def write_DCN_featureDict(df_all, params):
df = df_all.drop(*params["dropFeatures"]).drop(params['label'])
featureDict = {}
tc = 0
for colName in df.columns:
if colName in params["numericCols"]:
continue
else: # colName in categoryCols
uniqueFeature = df.select(
colName).distinct().toPandas()[colName].astype('float').values
featureDict[colName] = dict(
zip(uniqueFeature, range(tc,
len(uniqueFeature) + tc)))
tc = tc + len(uniqueFeature)
with open(params["featureDict_fileName"], 'wb') as f:
pickle.dump(featureDict, f)
logger.info("====\"{}\" finished ====".format(
params["featureDict_fileName"].split('/')[-1]))
utils.delete_before4_localData(params["featureDict_fileName"], params)
def globalAirport_data(params):
columns = ["Airport_code", "latitude", "longitude"]
with open(params["localFileName_org_Airport_data"], 'w') as f:
f.write(','.join(columns))
f.write('\n')
for sample in cursor2.find({}):
Airport_code = sample.get("_id")
latitude = sample.get("latitude")
longitude = sample.get("longitude")
try:
content = ','.join([Airport_code, latitude, longitude])
f.write(content + '\n')
except:
continue
logger.info("====\"{}\" finished====".format(
params["localFileName_org_Airport_data"].split('/')[-1]))
utils.delete_before4_localData(params["localFileName_org_Airport_data"],
params)
utils.upload_to_hdfs(params["localFileName_org_Airport_data"],
params["sparkDirName_org_Airport_data"], params)
def seatleft_data_add(params):
tomorrow_str2 = datetime.strftime(
params["generateDate"] + timedelta(days=1), '%Y-%m-%d')
today_monthDay = datetime.strftime(params["generateDate"], '%m-%d')
yesterday_str2 = (params["generateDate"] -
timedelta(days=1)).strftime('%Y-%m-%d')
after30_str2 = datetime.strftime(
params["generateDate"] + timedelta(days=30), '%Y-%m-%d')
monthDay_list = pd.date_range(
start=yesterday_str2, end=after30_str2,
freq='d').map(lambda x: datetime.strftime(x, '%m-%d')).to_list()
monthDay_list.remove(today_monthDay)
columns = ['queryDatetime', 'seatLeft', 'seatLeft_id']
with open(params["localFileName_org_seatLeft_data_add"], 'w') as f:
f.write(','.join(columns))
f.write('\n')
f.seek(2)
for monthDay in monthDay_list:
for sample in cursor3.find(
{'_id': {
"$regex": r'.*{}$'.format(monthDay)
}}):
seatLeft_id = sample.get('_id')
df = pd.DataFrame.from_dict(
sample.get('fc'),
orient='index').reset_index().rename(columns={
'index': 'queryDatetime',
0: 'seatLeft'
})
df['seatLeft_id'] = seatLeft_id
df.to_csv(params["localFileName_org_seatLeft_data_add"],
header=False,
index=False,
mode='a')
logger.info("====\"{}\" finished====".format(
params["localFileName_org_seatLeft_data_add"].split('/')[-1]))
utils.delete_before4_localData(
params["localFileName_org_seatLeft_data_add"], params)
utils.upload_to_hdfs(params["localFileName_org_seatLeft_data_add"],
params["sparkDirName_org_seatLeft_data_add"], params)
def lowPrice_train_data_add(params):
columns = ["queryDate", 'price', 'id', 'org', 'dst']
yesterday_monthDay_str = (params["generateDate"] -
timedelta(days=1)).strftime('%m-%d')
with open(params["localFileName_org_lowPrice_data_add"], 'w') as f:
f.write(','.join(columns))
f.write('\n')
f.seek(2)
for sample in cursor.find(
{'_id': {
"$regex": r'.*{}$'.format(yesterday_monthDay_str)
}}):
lowPrice_id = sample.get('_id')
del sample['_id']
org = sample.get('dairport')
del sample['dairport']
dst = sample.get('aairport')
del sample['aairport']
df = pd.DataFrame.from_dict(
sample, orient='index').reset_index().rename(columns={
'index': 'queryDate',
0: 'price'
})
df['id'] = lowPrice_id
df['org'] = org
df['dst'] = dst
df.to_csv(params["localFileName_org_lowPrice_data_add"],
header=False,
index=False,
mode='a')
logger.info("=====\"{}\" finished======".format(
params["localFileName_org_lowPrice_data_add"].split('/')[-1]))
utils.delete_before4_localData(
params["localFileName_org_lowPrice_data_add"], params)
utils.upload_to_hdfs(params["localFileName_org_lowPrice_data_add"],
params["sparkDirName_org_lowPrice_data_add"], params)