def get_actuals(cutoff_date_):
fdir = os.path.expanduser('~/my_tmp/cleaned/') # '~/my_tmp/in_df_data_'
adf = None
for f in os.listdir(fdir):
if str(
cutoff_date_.date()
) in f and 'tickets_' in f and 'old' not in f: # 'in_df_data_' in f: # we do not know the rolling window
s_ut.my_print('getting actuals from ' + fdir + f)
adf = p_ut.read_df(fdir + f)
break
if adf is None:
s_ut.my_print('no available actuals data for ' +
str(cutoff_date_.date()))
return None
adf.reset_index(inplace=True, drop=True)
p_ut.clean_cols(adf,
["language", "service_tier", "channel", "business_unit"],
'~/my_repos/capacity_planning/data/config/col_values.json',
check_new=False,
do_nan=False,
rename=True)
adf.rename(columns={
'ticket_count': 'y',
'ds_week_starting': 'ds'
},
inplace=True)
i_vals = ['nan', 'NULL', None, 'other', np.nan, 'null', 'N/A']
imp_data = imputer.impute(adf, i_vals=i_vals, ex_cols=['ds'])
imp_data['y'] = np.round(imp_data['y'].values, 0)
return imp_data
def tmp_ratios(cu, window, gcols): # tmp fix
wf = pd.read_parquet('~/my_tmp/cleaned/old_tickets_2020-02-29.par')
_ = p_ut.clean_cols(
wf, ["language", "service_tier", "channel", "business_unit"],
'~/my_repos/capacity_planning/data/config/col_values.json',
check_new=True,
do_nan=True,
rename=True)
wf.rename(columns={'ds_week_starting': 'ds'}, inplace=True)
wf['channel'] = wf.apply(
lambda x: 'directly'
if x['service_tier'] == 'directly' else x['channel'],
axis=1)
i_vals = ['nan', 'NULL', None, 'other', np.nan, 'null', 'N/A']
wf['ds'] = wf['ds'].dt.date.astype(str)
wf = imputer.impute(wf, i_vals=i_vals, ex_cols=['ds'])
wf['ds'] = pd.to_datetime(wf['ds'])
wf = wf[(wf['ds'] <= cu)
& (wf['ds'] >= cu - pd.to_timedelta(window, unit='W'))].copy()
wf['channel'] = wf.apply(
lambda x: 'directly'
if x['service_tier'] == 'directly' else x['channel'],
axis=1) # again in case imputation added directly wrongly
a_df = wf[wf['channel'] != 'directly'].copy()
lct_df = a_df.groupby(gcols).sum(numeric_only=True).reset_index()
l_df = lct_df.groupby(['language']).sum(numeric_only=True).reset_index()
lct_ratio = lct_df.merge(l_df, on=['language'], how='left')
lct_ratio['ratio'] = lct_ratio['ticket_count' +
'_x'] / lct_ratio['ticket_count' + '_y']
lct_ratio.drop(['ticket_count' + '_x', 'ticket_count' + '_y'],
axis=1,
inplace=True)
return lct_ratio
def __init__(self, file_path):
s_ut.my_print('setting forecast from ' + file_path)
t_info = file_path.split('.')[0].split('/')[-1]
self.raw = True if 'raw' in t_info else False
self.adj = not self.raw
self.rolling = True if '_r_' in t_info else False
self.cutoff_date = pd.to_datetime(t_info.split('_')[-1])
self.has_actuals = True if '_xls_' in t_info else False
self.data = p_ut.read_df(file_path)
p_ut.clean_cols(
self.data,
["language", "service_tier", "channel", "business_unit"],
'~/my_repos/capacity_planning/data/config/col_values.json',
check_new=False,
do_nan=False,
rename=True)
if 'ds_week_ending' in self.data.columns:
self.data['ds'] = pd.to_datetime(
self.data['ds_week_ending']) - pd.to_timedelta(6, unit='D')
self.data.drop('ds_week_ending', inplace=True, axis=1)
self.forecast = (self.cutoff_date +
pd.to_timedelta(7, unit='D')).month_name()
self.froot = file_path.split('.')[0][:-10]
]
dfx.columns = l_cols + dr
dfm = pd.melt(dfx,
value_vars=dr,
id_vars=l_cols,
var_name='ds_week_ending',
value_name='cx_yhat')
start = pd.to_datetime(cutoff_date) + pd.to_timedelta(
7, unit='D') # cutoff is a week_ending date. Go to the next week
dfm['ds_week_ending'] = pd.to_datetime(dfm['ds_week_ending'].values)
dfm = dfm[(dfm['ds_week_ending'] >= start)
& (dfm['ds_week_ending'] <= horizon_date)].copy()
_ = p_ut.clean_cols(
dfm, ['sector', 'language'],
'~/my_repos/capacity_planning/data/config/col_values.json',
check_new=False)
dfm['cx_yhat'] = dfm['cx_yhat'].apply(lambda x: to_float(x))
p_df = pd.pivot_table(dfm,
index=['ds_week_ending', 'language', 'sector'],
values='cx_yhat',
columns=['type']).reset_index()
c_df = p_df[p_df['sector'].isin([
'Claims', 'Community Education', 'Experiences', 'PST', 'Payments',
'Regulatory Response', 'Resolutions 1', 'Resolutions 2', 'Safety'
])].copy()
c_df.fillna(0, inplace=True)
# language level: no language level agg for FTE
if 'inbound-vol' in ts_name:
g_df = c_df.groupby(['ds_week_ending', 'language']).agg({
end = time.time()
s_ut.my_print('imputer: reduce secs: ' + str(end - start))
return z_all
if __name__ == '__main__':
# df = pd.DataFrame({
# 'a': [1,2,np.nan, 4, 5, 6, 7],
# 'b': ['x', 'y', None, None, 'z', 'x', 'z'],
# 'c': [1.5, 2.3, 5.2, 3, None, np.nan, 5.2],
# 'ds': ['Mon', 'Tues', 'Wed', 'Thurs', 'Fri', 'Sat', 'Sun'],
# 'd': [None, None, 'a', 'bb', 'a', 'a', 'bb']}
# )
# ex_cols_ = ['ds']
df = pd.read_parquet('~/my_tmp/phone-aht_2019-11-23.par')
from capacity_planning.utilities import pandas_utils as p_ut
new_vals = p_ut.clean_cols(
df, ["service_region", "language", "sector", "interaction_type"],
'~/my_repos/capacity_planning/data/config/col_values.json',
check_new=True,
do_nan=True)
ex_cols_ = ['ds', 'agent_id']
df['tenure_days'] = df['tenure_days'].apply(lambda x: x
if x > 0 else np.nan)
# df = df[df['ds'] >= '2019-06-01'].copy()
print(df.head(10))
zz = impute(df, ex_cols=ex_cols_)
print(zz.head(10))
print('DONE')
# avg_fdf['initiative'] = False
# s_ut.my_print('saving raw fcast data to ' + fcast_f)
# p_ut.save_df(avg_fdf, fcast_f) # ONLY forecasted data
# accuracy for the <months> months old raw forecast
adf = errs.get_actuals(cutoff_date) # raw actuals up to cutoff_date
window = 4
fcast_file = errs.get_fcast_file(
cutoff_date,
'~/Forecasts/rolling/par/raw_r_fcast_' + str(window) + '_',
months=3) # file path from <months> months old forecast from cutoff
fdf_obj = ts.TicketForecast(fcast_file) # fcast obj from 3 months ago
fdf = fdf_obj.data
p_ut.clean_cols(fdf,
["language", "service_tier", "channel", "business_unit"],
'~/my_repos/capacity_planning/data/config/col_values.json',
check_new=False,
do_nan=False,
rename=True)
fdf.rename(columns={'ticket_count': 'forecasted_count'}, inplace=True)
# fdf['ds_week_ending'] = pd.to_datetime(fdf['ds_week_ending'])
# fdf['ds'] = fdf['ds_week_ending'] - pd.to_timedelta(6, unit='D')
if fdf is None or adf is None:
lang_errs, tier_errs = None, None
else:
s_ut.my_print('Error wrt actuals for an old forecast')
lang_errs, tier_errs, off_df = errs.get_errs(
cutoff_date, fdf_obj, adf,
tcol='ds') # errs on filtered actuals for old raw fcast
lang_errs['adj'] = False
lang_errs['initiative'] = False
tier_errs['adj'] = False