class PlotlyAPI(object):
def __init__(self):
self.influx = InfluxDB()
def _create_scatter(self,expression,mode):
try:
axis_separate = expression.index(':')
axis_label = expression[:axis_separate]
expression = expression[axis_separate+1:]
except:
axis_label = 'y1'
try:
name_separate = expression.index(';')
legend = expression[name_separate+1:]
except:
name_separate = len(expression)
legend = expression
result = self.influx.interpret(expression[:name_separate])
x = result.index.to_pydatetime()
y = result['value'].get_values()
return Scatter(x=x,y=y,mode=mode,name=legend)
def _create_ts(self,exp_list,title,mode='lines'):
scatters = []
for exp in exp_list:
scatters.append(self._create_scatter(exp,mode))
data = Data(scatters)
py.plot(data, filename=title)
processed_list = [
[ 'wage_lag_15m', 'mlag(wage_us_yoy,15)' ],
[ 'zillow_median_sale_price_15m', 'mlag(zillow_median_sale_price_yoy,15)' ],
]
df = DataFrameClient('localhost',8086,'root','root')
if({'name':'Econ'} not in df.get_list_database()):
df.create_database('Econ')
df.switch_database('Econ')
#add all items in interested list
start = timeit.default_timer()
for item in interest_list:
if item[0] in df.get_list_series():
df.delete_series(item[0])
results = influx_Fred.interpret(item[1])
results = results.replace(to_replace='NaN',value='.')
results = DataFrame({'value':results['value']})
df.write_points({item[0]:results})
print item
print 'total time in seconds: %.2f' % (timeit.default_timer() - start)
#add all items in processed list
start = timeit.default_timer()
influx_Econ = InfluxDB(db_name='Econ')
for item in processed_list:
if item[0] in df.get_list_series():
df.delete_series(item[0])
results = influx_Econ.interpret(item[1])
results = results.replace(to_replace='NaN',value='.')
results = DataFrame({'value':results['value']})
'''
Created on Jul 26, 2015
@author: shaunz
'''
from QueryData.InfluxQuery import InfluxDB
from pandas.core.config import option_context
influx = InfluxDB()
one = influx.query('cn_gdp_cum')
two = influx.interpret('anticum(cn_gdp_cum,Q)*100')
with option_context('display.max_rows', 9999, 'display.max_columns', 3):
print one
print two
'''
new_series = {}
for each in one.iterrows():
if each[0].month != 1:
prev_date = each[0] + DateOffset(months=-1)
prev_date = prev_date.to_period('M').to_timestamp('M')
monthly = each[1]['value']-one.loc[prev_date]
new_series[each[0]] = monthly['value']
else:
new_series[each[0]] = each[1]['value']
Series(new_series)
'''
'''
Created on Jul 18, 2015
@author: shaunz
'''
from QueryData.InfluxQuery import InfluxDB
from pandas.stats.moments import rolling_corr
from pandas.tools.merge import concat
from pandas.core.common import isnull
from pandas.core.config import option_context
db = InfluxDB(db_name='Econ')
series_1 = db.interpret('cpi_us_owners_equiv_yoy')
series_2 = db.interpret('avg(zillow_median_sale_price_yoy,M)')
series_1.rename(columns={'value':'value_1'}, inplace=True)
series_2.rename(columns={'value':'value_2'}, inplace=True)
periods = 60
analysis_type = 'snapshot'
if(analysis_type=='snapshot'):
for i in range(-36, 37): # go back i periods
shifted_1 = series_1.shift(i)
combined = concat([shifted_1['value_1'],series_2['value_2']],axis=1,join='outer')
combined = combined.loc[isnull(combined['value_1']) != True]
combined = combined.loc[isnull(combined['value_2']) != True]
results = rolling_corr(combined['value_1'],combined['value_2'],window=periods)
print '%d,%.4f' % (i, results[-1:].values)
elif(analysis_type=='history'):
shifted_1 = series_1.shift(0)
combined = concat([shifted_1['value_1'],series_2['value_2']],axis=1,join='outer')
combined = combined.loc[isnull(combined['value_1']) != True]