In [1]:

    

# remove the notebook root logger.\n",
import logging
logger = logging.getLogger()
logger.handlers = []


    

In [2]:

    

import pandas as pd


    

In [3]:

    

import datetime

filename = "https://raw.githubusercontent.com/antoinecarme/TimeSeriesData/master/Stat_FR_CommExt/merged/French_Wine_Export_in_Euros_Some_Countries.csv"

French_Wine_Export_in_Euros_DF = pd.read_csv(filename);

lDateColumn = 'Month';
French_Wine_Export_in_Euros_DF[lDateColumn] = French_Wine_Export_in_Euros_DF[lDateColumn].apply(lambda x : datetime.datetime.strptime(str(x), "%Y-%m-%d"))

French_Wine_Export_in_Euros_DF.head(5)


    

    
Out[3]:







  

    

      

      
Month
      
ALSACE_BLANC_BE
      
ALSACE_BLANC_CA
      
ALSACE_BLANC_CH
      
ALSACE_BLANC_CN
      
ALSACE_BLANC_DE
      
ALSACE_BLANC_GB
      
ALSACE_BLANC_HK
      
ALSACE_BLANC_JP
      
ALSACE_BLANC_NL
      
...
      
RHÔNE_ROUGE_CA
      
RHÔNE_ROUGE_CH
      
RHÔNE_ROUGE_CN
      
RHÔNE_ROUGE_DE
      
RHÔNE_ROUGE_GB
      
RHÔNE_ROUGE_HK
      
RHÔNE_ROUGE_JP
      
RHÔNE_ROUGE_NL
      
RHÔNE_ROUGE_OTHER
      
RHÔNE_ROUGE_US
    
  
  

    

      
0
      
2012-01-01
      
1131852
      
571940
      
257116
      
140013
      
357516
      
344274
      
51043
      
329883
      
624695
      
...
      
800628
      
650236
      
1127393
      
783952
      
1513332
      
335710
      
916965
      
314654
      
35418
      
1260510
    
    

      
1
      
2012-02-01
      
857289
      
585556
      
187917
      
89805
      
505896
      
424953
      
80220
      
239242
      
899144
      
...
      
795356
      
680567
      
188827
      
868576
      
1277664
      
197072
      
318522
      
516140
      
36604
      
2136484
    
    

      
2
      
2012-03-01
      
1438501
      
834142
      
679281
      
103577
      
583887
      
352675
      
102980
      
394945
      
1093638
      
...
      
873858
      
1318538
      
856382
      
627236
      
2343448
      
132200
      
605914
      
410308
      
43197
      
3264810
    
    

      
3
      
2012-04-01
      
986602
      
866221
      
523202
      
203456
      
515903
      
403517
      
52773
      
260953
      
874217
      
...
      
1004028
      
919493
      
460815
      
829432
      
1861188
      
128164
      
472522
      
290407
      
45396
      
2631466
    
    

      
4
      
2012-05-01
      
991668
      
1061702
      
611369
      
80327
      
767522
      
627232
      
78115
      
680323
      
994697
      
...
      
2098300
      
632778
      
239200
      
934532
      
2215922
      
202210
      
337894
      
284868
      
41469
      
2945175
    
  

5 rows × 177 columns

In [4]:

    

French_Wine_Export_in_Euros_DF.describe()


    

    
Out[4]:







  

    

      

      
ALSACE_BLANC_BE
      
ALSACE_BLANC_CA
      
ALSACE_BLANC_CH
      
ALSACE_BLANC_CN
      
ALSACE_BLANC_DE
      
ALSACE_BLANC_GB
      
ALSACE_BLANC_HK
      
ALSACE_BLANC_JP
      
ALSACE_BLANC_NL
      
ALSACE_BLANC_OTHER
      
...
      
RHÔNE_ROUGE_CA
      
RHÔNE_ROUGE_CH
      
RHÔNE_ROUGE_CN
      
RHÔNE_ROUGE_DE
      
RHÔNE_ROUGE_GB
      
RHÔNE_ROUGE_HK
      
RHÔNE_ROUGE_JP
      
RHÔNE_ROUGE_NL
      
RHÔNE_ROUGE_OTHER
      
RHÔNE_ROUGE_US
    
  
  

    

      
count
      
5.000000e+01
      
5.000000e+01
      
50.000000
      
50.000000
      
5.000000e+01
      
50.000000
      
50.000000
      
50.000000
      
5.000000e+01
      
50.000000
      
...
      
5.000000e+01
      
5.000000e+01
      
5.000000e+01
      
5.000000e+01
      
5.000000e+01
      
50.000000
      
5.000000e+01
      
5.000000e+01
      
50.000000
      
5.000000e+01
    
    

      
mean
      
1.189956e+06
      
8.037735e+05
      
428923.380000
      
119348.360000
      
6.173951e+05
      
414037.100000
      
88636.820000
      
361437.900000
      
8.035454e+05
      
38337.801569
      
...
      
1.607702e+06
      
8.523710e+05
      
7.498312e+05
      
1.072562e+06
      
2.876061e+06
      
235678.640000
      
5.760031e+05
      
4.986103e+05
      
52907.080000
      
2.912355e+06
    
    

      
std
      
2.780384e+05
      
1.691507e+05
      
158179.648178
      
65116.609016
      
2.765474e+05
      
116398.158498
      
68347.582681
      
110140.042643
      
2.732401e+05
      
6088.988573
      
...
      
7.064099e+05
      
2.458533e+05
      
3.592653e+05
      
4.307817e+05
      
1.366642e+06
      
140071.773582
      
2.398091e+05
      
1.720202e+05
      
9667.885799
      
6.675896e+05
    
    

      
min
      
5.477480e+05
      
3.514620e+05
      
88329.000000
      
19546.000000
      
3.131500e+05
      
168388.000000
      
31864.000000
      
192553.000000
      
3.085880e+05
      
18550.156250
      
...
      
7.935600e+05
      
3.992940e+05
      
1.888270e+05
      
5.558550e+05
      
1.277664e+06
      
77550.000000
      
2.053840e+05
      
2.311020e+05
      
35238.000000
      
1.260510e+06
    
    

      
25%
      
1.004816e+06
      
6.791068e+05
      
316753.500000
      
82278.500000
      
4.801042e+05
      
333264.500000
      
52845.750000
      
293690.250000
      
6.061862e+05
      
35457.417067
      
...
      
1.047350e+06
      
6.819278e+05
      
5.089532e+05
      
7.219495e+05
      
2.025437e+06
      
152307.750000
      
3.751060e+05
      
3.725818e+05
      
46721.000000
      
2.500847e+06
    
    

      
50%
      
1.161726e+06
      
8.070940e+05
      
445554.500000
      
98656.000000
      
5.449660e+05
      
401774.500000
      
76561.500000
      
335180.000000
      
7.884350e+05
      
38715.394123
      
...
      
1.506712e+06
      
8.120535e+05
      
7.065070e+05
      
9.246160e+05
      
2.519672e+06
      
203395.000000
      
5.414685e+05
      
4.674045e+05
      
52305.000000
      
2.896628e+06
    
    

      
75%
      
1.323582e+06
      
8.880218e+05
      
542924.000000
      
161694.000000
      
6.549058e+05
      
515408.000000
      
102011.750000
      
417475.000000
      
8.980862e+05
      
41900.001008
      
...
      
1.936743e+06
      
1.023269e+06
      
9.285735e+05
      
1.207832e+06
      
3.018630e+06
      
257873.500000
      
7.998980e+05
      
6.143272e+05
      
59756.250000
      
3.380802e+06
    
    

      
max
      
2.166585e+06
      
1.216720e+06
      
857207.000000
      
317661.000000
      
1.871515e+06
      
635447.000000
      
510151.000000
      
680323.000000
      
1.458234e+06
      
52823.280702
      
...
      
3.946934e+06
      
1.334550e+06
      
1.755449e+06
      
2.312776e+06
      
8.652432e+06
      
879037.000000
      
1.076535e+06
      
1.063568e+06
      
71487.000000
      
4.664273e+06
    
  

8 rows × 176 columns

In [5]:

    

%matplotlib inline


    

In [6]:

    

French_Wine_Export_in_Euros_DF.plot('Month' , 
                                    ['CHAMPAGNE_MOUSSEUX_DE', 'CHAMPAGNE_MOUSSEUX_GB' , 'CHAMPAGNE_MOUSSEUX_JP'],
                                   figsize=(32 , 12)).legend(loc='best', fontsize=24)


    

    
Out[6]:





<matplotlib.legend.Legend at 0x7feb3960b730>






    

In [7]:

    

French_Wine_Export_in_Euros_DF.plot('Month' , 
                                    ['BORDEAUX_BLANC_US', 'BORDEAUX_ROUGE_US'],
                                    fontsize = 24,
                                    figsize=(32 , 12)).legend(loc='best', fontsize=24)


    

    
Out[7]:





<matplotlib.legend.Legend at 0x7feb395590d0>






    

In [8]:

    

CN_columns = [col for col in French_Wine_Export_in_Euros_DF.columns if col.endswith('_CN') ]
French_Wine_Export_in_Euros_DF[CN_columns].describe()


    

    
Out[8]:







  

    

      

      
ALSACE_BLANC_CN
      
BEAUJOLAIS_ROUGE_CN
      
BORDEAUX_BLANC_CN
      
BORDEAUX_ROUGE_CN
      
BOURGOGNE_BLANC_CN
      
BOURGOGNE_ROUGE_CN
      
CHAMPAGNE_MOUSSEUX_CN
      
EUROPE_ROUGE_CN
      
FRANCE_BLANC_CN
      
FRANCE_MOUSSEUX_CN
      
FRANCE_ROUGE_CN
      
LANGUEDOC_ROUGE_CN
      
LOIRE_BLANC_CN
      
LOIRE_ROUGE_CN
      
OTHER_ROUGE_CN
      
RHÔNE_ROUGE_CN
    
  
  

    

      
count
      
50.000000
      
50.00000
      
5.000000e+01
      
5.000000e+01
      
50.000000
      
5.000000e+01
      
5.000000e+01
      
5.000000e+01
      
50.000000
      
50.000000
      
50.000000
      
5.000000e+01
      
50.000000
      
50.000000
      
50.000000
      
5.000000e+01
    
    

      
mean
      
119348.360000
      
130292.47000
      
4.993186e+05
      
1.138432e+07
      
242575.340000
      
7.436432e+05
      
6.225160e+05
      
8.313372e+05
      
12978.062000
      
93298.256667
      
11185.281403
      
1.304245e+06
      
48771.330000
      
92016.740000
      
52969.948000
      
7.498312e+05
    
    

      
std
      
65116.609016
      
102214.45714
      
2.936971e+05
      
3.081051e+06
      
99492.736867
      
4.176045e+05
      
3.812781e+05
      
2.337996e+05
      
24705.479992
      
63521.773261
      
13888.069451
      
6.352887e+05
      
27922.200988
      
54704.609614
      
36129.014739
      
3.592653e+05
    
    

      
min
      
19546.000000
      
5833.00000
      
1.678740e+05
      
5.377235e+06
      
75396.000000
      
2.725460e+05
      
1.976760e+05
      
3.270090e+05
      
0.000000
      
20488.500000
      
260.500000
      
4.916860e+05
      
16643.500000
      
15325.000000
      
6337.750000
      
1.888270e+05
    
    

      
25%
      
82278.500000
      
66206.50000
      
3.117738e+05
      
9.737536e+06
      
171633.250000
      
4.214805e+05
      
3.897600e+05
      
6.529608e+05
      
1607.250000
      
46355.750000
      
4759.017857
      
9.207640e+05
      
31191.250000
      
54588.250000
      
28779.650000
      
5.089532e+05
    
    

      
50%
      
98656.000000
      
98427.50000
      
4.148980e+05
      
1.039686e+07
      
237099.500000
      
6.060540e+05
      
5.299100e+05
      
8.078685e+05
      
6237.750000
      
82644.583333
      
8837.333333
      
1.211934e+06
      
37845.750000
      
77491.500000
      
44118.875000
      
7.065070e+05
    
    

      
75%
      
161694.000000
      
158673.75000
      
5.623658e+05
      
1.316721e+07
      
308119.750000
      
9.648268e+05
      
7.029695e+05
      
9.859902e+05
      
14842.000000
      
118693.250000
      
12869.584821
      
1.416010e+06
      
62140.375000
      
119915.500000
      
62176.000000
      
9.285735e+05
    
    

      
max
      
317661.000000
      
453985.00000
      
1.890312e+06
      
1.905088e+07
      
535157.000000
      
2.021852e+06
      
2.103130e+06
      
1.367882e+06
      
163701.000000
      
366064.500000
      
96564.222222
      
3.607517e+06
      
151128.000000
      
272948.000000
      
179114.750000
      
1.755449e+06
    
  

In [9]:

    

French_Wine_Export_in_Euros_DF.plot('Month' , 
                                    CN_columns,
                                   figsize=(32 , 12)).legend(fontsize=18)


    

    
Out[9]:





<matplotlib.legend.Legend at 0x7feb33b75a00>






    

In [10]:

    

Wines = "ALSACE BEAUJOLAIS BORDEAUX BOURGOGNE CHAMPAGNE EUROPE FRANCE LANGUEDOC LOIRE OTHER RHÔNE".split();
Variants = ['BLANC' , 'MOUSSEUX' , 'ROUGE'];
Countries = ['GB', 'US', 'DE', 'BE', 'CN', 'JP', 'CH', 'HK', 'NL', 'CA' , 'OTHER']


    

In [11]:

    

# A lot of possibilities , but only 176 are available ;)
len(Wines) * len(Variants) * len(Countries)


    

    
Out[11]:





363

In [12]:

    

from IPython.display import display


    

In [13]:

    

rows = [];
cols = []
for cntry in Countries:
    cntry_columns = [col for col in French_Wine_Export_in_Euros_DF.columns if col.endswith('_' + cntry) ]
    cols = []
    row = [cntry];
    for col in cntry_columns:
        col1 = col.replace('_' + cntry , "")
        sum1 = French_Wine_Export_in_Euros_DF[col].sum()
        row.append(sum1);
        cols.append(col1);
    rows.append(row);

plot_df = pd.DataFrame(rows , columns=['Country'] + cols);

    
plot_df.set_index('Country').plot.barh(stacked=True, figsize=(20,10), fontsize = 16, colormap='Paired').legend(loc='best', fontsize=12)


    

    
Out[13]:





<matplotlib.legend.Legend at 0x7feb33b316d0>

In [ ]:

    




    

In [14]:

    

Regions = ['EUROPE', 'AMERICA', 'EUROPE' , 'EUROPE' , 'ASIA' , 'ASIA' , 'EUROPE',  'ASIA', 'EUROPE' , 'AMERICA' , 'OTHER_REGION']
lDict = dict(zip(Countries , Regions));

# simplify !!!!
Variants = ['BLANC' , 'ROUGE'];
Wines = Wines[0:3];
Countries = Countries[0:5]


    

In [15]:

    

# Hierarchy defintion
rows_list = [];
for v in Variants:
    for w in Wines:
        for c in Countries:
            col = w + "_" + v + "_" + c;
            region = lDict[c]
            if(col in French_Wine_Export_in_Euros_DF.columns):
                rows_list.append([col , c , region , 'WORLD']);
            
lLevels = ['Wine' , 'Country' , 'Region' , 'WORLD'];
lHierarchy = {};
lHierarchy['Levels'] = lLevels;
lHierarchy['Data'] = pd.DataFrame(rows_list, columns =  lLevels);
lHierarchy['Type'] = "Hierarchical";


    

In [16]:

    

lHierarchy


    

    
Out[16]:





{'Levels': ['Wine', 'Country', 'Region', 'WORLD'],
 'Data':                    Wine Country   Region  WORLD
 0       ALSACE_BLANC_GB      GB   EUROPE  WORLD
 1       ALSACE_BLANC_US      US  AMERICA  WORLD
 2       ALSACE_BLANC_DE      DE   EUROPE  WORLD
 3       ALSACE_BLANC_BE      BE   EUROPE  WORLD
 4       ALSACE_BLANC_CN      CN     ASIA  WORLD
 5     BORDEAUX_BLANC_GB      GB   EUROPE  WORLD
 6     BORDEAUX_BLANC_US      US  AMERICA  WORLD
 7     BORDEAUX_BLANC_DE      DE   EUROPE  WORLD
 8     BORDEAUX_BLANC_BE      BE   EUROPE  WORLD
 9     BORDEAUX_BLANC_CN      CN     ASIA  WORLD
 10  BEAUJOLAIS_ROUGE_GB      GB   EUROPE  WORLD
 11  BEAUJOLAIS_ROUGE_US      US  AMERICA  WORLD
 12  BEAUJOLAIS_ROUGE_DE      DE   EUROPE  WORLD
 13  BEAUJOLAIS_ROUGE_BE      BE   EUROPE  WORLD
 14  BEAUJOLAIS_ROUGE_CN      CN     ASIA  WORLD
 15    BORDEAUX_ROUGE_GB      GB   EUROPE  WORLD
 16    BORDEAUX_ROUGE_US      US  AMERICA  WORLD
 17    BORDEAUX_ROUGE_DE      DE   EUROPE  WORLD
 18    BORDEAUX_ROUGE_BE      BE   EUROPE  WORLD
 19    BORDEAUX_ROUGE_CN      CN     ASIA  WORLD,
 'Type': 'Hierarchical'}

In [ ]:

    




    

In [17]:

    

# create a model to plot the hierarchy.
import pyaf.HierarchicalForecastEngine as hautof
lEngine = hautof.cHierarchicalForecastEngine()


lSignalHierarchy = lEngine.plot_Hierarchy(French_Wine_Export_in_Euros_DF , "Month", "Signal", 1, 
                                          lHierarchy, None);


    

    




INFO:pyaf.std:START_HIERARCHICAL_PLOTTING






    













    




INFO:pyaf.std:END_HIERARCHICAL_PLOTTING_TIME_IN_SECONDS 0.18779993057250977

In [18]:

    

lSignalHierarchy.mStructure


    

    
Out[18]:





{0: {'ALSACE_BLANC_GB': set(),
  'ALSACE_BLANC_US': set(),
  'ALSACE_BLANC_DE': set(),
  'ALSACE_BLANC_BE': set(),
  'ALSACE_BLANC_CN': set(),
  'BORDEAUX_BLANC_GB': set(),
  'BORDEAUX_BLANC_US': set(),
  'BORDEAUX_BLANC_DE': set(),
  'BORDEAUX_BLANC_BE': set(),
  'BORDEAUX_BLANC_CN': set(),
  'BEAUJOLAIS_ROUGE_GB': set(),
  'BEAUJOLAIS_ROUGE_US': set(),
  'BEAUJOLAIS_ROUGE_DE': set(),
  'BEAUJOLAIS_ROUGE_BE': set(),
  'BEAUJOLAIS_ROUGE_CN': set(),
  'BORDEAUX_ROUGE_GB': set(),
  'BORDEAUX_ROUGE_US': set(),
  'BORDEAUX_ROUGE_DE': set(),
  'BORDEAUX_ROUGE_BE': set(),
  'BORDEAUX_ROUGE_CN': set()},
 1: {'GB': {'ALSACE_BLANC_GB',
   'BEAUJOLAIS_ROUGE_GB',
   'BORDEAUX_BLANC_GB',
   'BORDEAUX_ROUGE_GB'},
  'US': {'ALSACE_BLANC_US',
   'BEAUJOLAIS_ROUGE_US',
   'BORDEAUX_BLANC_US',
   'BORDEAUX_ROUGE_US'},
  'DE': {'ALSACE_BLANC_DE',
   'BEAUJOLAIS_ROUGE_DE',
   'BORDEAUX_BLANC_DE',
   'BORDEAUX_ROUGE_DE'},
  'BE': {'ALSACE_BLANC_BE',
   'BEAUJOLAIS_ROUGE_BE',
   'BORDEAUX_BLANC_BE',
   'BORDEAUX_ROUGE_BE'},
  'CN': {'ALSACE_BLANC_CN',
   'BEAUJOLAIS_ROUGE_CN',
   'BORDEAUX_BLANC_CN',
   'BORDEAUX_ROUGE_CN'}},
 2: {'EUROPE': {'BE', 'DE', 'GB'}, 'AMERICA': {'US'}, 'ASIA': {'CN'}},
 3: {'WORLD': {'AMERICA', 'ASIA', 'EUROPE'}}}

In [ ]:

    




    

In [ ]:

    




    

In [ ]:

    




    

In [19]:

    

# create a hierarchical model and train it
import pyaf.HierarchicalForecastEngine as hautof

lEngine = hautof.cHierarchicalForecastEngine()

lSignalVar = "Sales";

#
N = French_Wine_Export_in_Euros_DF.shape[0];
H = 4;
train_df = French_Wine_Export_in_Euros_DF.head(N-H);


lSignalHierarchy = lEngine.train(train_df , lDateColumn, lSignalVar, 1, lHierarchy, None);


    

    




INFO:pyaf.std:START_HIERARCHICAL_TRAINING
INFO:pyaf.hierarchical:TRAINING_HIERARCHICAL_MODELS_LEVEL_SIGNAL [(0, 'ALSACE_BLANC_BE'), (0, 'ALSACE_BLANC_CN'), (0, 'ALSACE_BLANC_DE'), (0, 'ALSACE_BLANC_GB'), (0, 'ALSACE_BLANC_US'), (0, 'BEAUJOLAIS_ROUGE_BE'), (0, 'BEAUJOLAIS_ROUGE_CN'), (0, 'BEAUJOLAIS_ROUGE_DE'), (0, 'BEAUJOLAIS_ROUGE_GB'), (0, 'BEAUJOLAIS_ROUGE_US'), (0, 'BORDEAUX_BLANC_BE'), (0, 'BORDEAUX_BLANC_CN'), (0, 'BORDEAUX_BLANC_DE'), (0, 'BORDEAUX_BLANC_GB'), (0, 'BORDEAUX_BLANC_US'), (0, 'BORDEAUX_ROUGE_BE'), (0, 'BORDEAUX_ROUGE_CN'), (0, 'BORDEAUX_ROUGE_DE'), (0, 'BORDEAUX_ROUGE_GB'), (0, 'BORDEAUX_ROUGE_US'), (1, 'BE'), (1, 'CN'), (1, 'DE'), (1, 'GB'), (1, 'US'), (2, 'AMERICA'), (2, 'ASIA'), (2, 'EUROPE'), (3, 'WORLD')]
INFO:pyaf.std:START_TRAINING 'ALSACE_BLANC_CN'
INFO:pyaf.std:START_TRAINING 'ALSACE_BLANC_BE'
INFO:pyaf.std:START_TRAINING 'ALSACE_BLANC_DE'
INFO:pyaf.std:START_TRAINING 'ALSACE_BLANC_GB'
INFO:pyaf.std:START_TRAINING 'ALSACE_BLANC_US'
INFO:pyaf.std:START_TRAINING 'BEAUJOLAIS_ROUGE_CN'
INFO:pyaf.std:START_TRAINING 'BEAUJOLAIS_ROUGE_BE'
INFO:pyaf.std:START_TRAINING 'BEAUJOLAIS_ROUGE_DE'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BEAUJOLAIS_ROUGE_DE' 3.116830348968506
INFO:pyaf.std:START_TRAINING 'BEAUJOLAIS_ROUGE_GB'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BEAUJOLAIS_ROUGE_CN' 3.2174572944641113
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BEAUJOLAIS_ROUGE_BE' 3.218780755996704
INFO:pyaf.std:START_TRAINING 'BEAUJOLAIS_ROUGE_US'
INFO:pyaf.std:START_TRAINING 'BORDEAUX_BLANC_BE'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'ALSACE_BLANC_CN' 3.2586417198181152
INFO:pyaf.std:START_TRAINING 'BORDEAUX_BLANC_CN'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'ALSACE_BLANC_DE' 3.2711479663848877
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'ALSACE_BLANC_GB' 3.280301809310913
INFO:pyaf.std:START_TRAINING 'BORDEAUX_BLANC_DE'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'ALSACE_BLANC_BE' 3.299868106842041
INFO:pyaf.std:START_TRAINING 'BORDEAUX_BLANC_GB'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'ALSACE_BLANC_US' 3.316903591156006
INFO:pyaf.std:START_TRAINING 'BORDEAUX_BLANC_US'
INFO:pyaf.std:START_TRAINING 'BORDEAUX_ROUGE_BE'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BEAUJOLAIS_ROUGE_GB' 2.790262460708618
INFO:pyaf.std:START_TRAINING 'BORDEAUX_ROUGE_CN'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BORDEAUX_BLANC_DE' 2.77897310256958
INFO:pyaf.std:START_TRAINING 'BORDEAUX_ROUGE_DE'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BEAUJOLAIS_ROUGE_US' 2.8733577728271484
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BORDEAUX_BLANC_CN' 2.856806993484497
INFO:pyaf.std:START_TRAINING 'BORDEAUX_ROUGE_GB'
INFO:pyaf.std:START_TRAINING 'BORDEAUX_ROUGE_US'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BORDEAUX_BLANC_GB' 2.8904035091400146
INFO:pyaf.std:START_TRAINING 'BE'
INFO:pyaf.std:START_TRAINING 'GB'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BORDEAUX_BLANC_BE' 2.9640254974365234
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BORDEAUX_BLANC_US' 2.930906295776367
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BORDEAUX_ROUGE_BE' 2.941695213317871
INFO:pyaf.std:START_TRAINING 'CN'
INFO:pyaf.std:START_TRAINING 'DE'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BORDEAUX_ROUGE_CN' 2.878282308578491
INFO:pyaf.std:START_TRAINING 'US'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BORDEAUX_ROUGE_DE' 2.7534682750701904
INFO:pyaf.std:START_TRAINING 'AMERICA'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BORDEAUX_ROUGE_GB' 2.8161003589630127
INFO:pyaf.std:START_TRAINING 'ASIA'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BORDEAUX_ROUGE_US' 2.8470776081085205
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'BE' 2.799729108810425
INFO:pyaf.std:START_TRAINING 'EUROPE'
INFO:pyaf.std:START_TRAINING 'WORLD'
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'DE' 2.83880877494812
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'CN' 2.852757692337036
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'GB' 2.96859073638916
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'US' 2.7348833084106445
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'ASIA' 2.6849465370178223
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'EUROPE' 2.679492712020874
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'AMERICA' 2.8715567588806152
INFO:pyaf.std:END_TRAINING_TIME_IN_SECONDS 'WORLD' 2.790771961212158
INFO:pyaf.hierarchical:TRAINING_HIERARCHICAL_MODEL_COMPUTE_TOP_DOWN_HISTORICAL_PROPORTIONS
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODELS_LEVEL_SIGNAL [(0, 'ALSACE_BLANC_BE'), (0, 'ALSACE_BLANC_CN'), (0, 'ALSACE_BLANC_DE'), (0, 'ALSACE_BLANC_GB'), (0, 'ALSACE_BLANC_US'), (0, 'BEAUJOLAIS_ROUGE_BE'), (0, 'BEAUJOLAIS_ROUGE_CN'), (0, 'BEAUJOLAIS_ROUGE_DE'), (0, 'BEAUJOLAIS_ROUGE_GB'), (0, 'BEAUJOLAIS_ROUGE_US'), (0, 'BORDEAUX_BLANC_BE'), (0, 'BORDEAUX_BLANC_CN'), (0, 'BORDEAUX_BLANC_DE'), (0, 'BORDEAUX_BLANC_GB'), (0, 'BORDEAUX_BLANC_US'), (0, 'BORDEAUX_ROUGE_BE'), (0, 'BORDEAUX_ROUGE_CN'), (0, 'BORDEAUX_ROUGE_DE'), (0, 'BORDEAUX_ROUGE_GB'), (0, 'BORDEAUX_ROUGE_US'), (1, 'BE'), (1, 'CN'), (1, 'DE'), (1, 'GB'), (1, 'US'), (2, 'AMERICA'), (2, 'ASIA'), (2, 'EUROPE'), (3, 'WORLD')]
INFO:pyaf.std:START_FORECASTING 'ALSACE_BLANC_BE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ALSACE_BLANC_BE' 0.03546404838562012
INFO:pyaf.std:START_FORECASTING 'ALSACE_BLANC_CN'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ALSACE_BLANC_CN' 0.046620845794677734
INFO:pyaf.std:START_FORECASTING 'ALSACE_BLANC_DE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ALSACE_BLANC_DE' 0.03574967384338379
INFO:pyaf.std:START_FORECASTING 'ALSACE_BLANC_GB'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ALSACE_BLANC_GB' 0.03654885292053223
INFO:pyaf.std:START_FORECASTING 'ALSACE_BLANC_US'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ALSACE_BLANC_US' 0.042937517166137695
INFO:pyaf.std:START_FORECASTING 'BEAUJOLAIS_ROUGE_BE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BEAUJOLAIS_ROUGE_BE' 0.04005265235900879
INFO:pyaf.std:START_FORECASTING 'BEAUJOLAIS_ROUGE_CN'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BEAUJOLAIS_ROUGE_CN' 0.03484606742858887
INFO:pyaf.std:START_FORECASTING 'BEAUJOLAIS_ROUGE_DE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BEAUJOLAIS_ROUGE_DE' 0.04229998588562012
INFO:pyaf.std:START_FORECASTING 'BEAUJOLAIS_ROUGE_GB'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BEAUJOLAIS_ROUGE_GB' 0.031418800354003906
INFO:pyaf.std:START_FORECASTING 'BEAUJOLAIS_ROUGE_US'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BEAUJOLAIS_ROUGE_US' 0.033751487731933594
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_BLANC_BE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_BLANC_BE' 0.04135394096374512
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_BLANC_CN'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_BLANC_CN' 0.02912306785583496
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_BLANC_DE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_BLANC_DE' 0.03085160255432129
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_BLANC_GB'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_BLANC_GB' 0.03562617301940918
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_BLANC_US'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_BLANC_US' 0.029346466064453125
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_ROUGE_BE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_ROUGE_BE' 0.03003406524658203
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_ROUGE_CN'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_ROUGE_CN' 0.032271623611450195
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_ROUGE_DE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_ROUGE_DE' 0.029767751693725586
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_ROUGE_GB'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_ROUGE_GB' 0.02918076515197754
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_ROUGE_US'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_ROUGE_US' 0.0391845703125
INFO:pyaf.std:START_FORECASTING 'BE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BE' 0.03859758377075195
INFO:pyaf.std:START_FORECASTING 'CN'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'CN' 0.03562211990356445
INFO:pyaf.std:START_FORECASTING 'DE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'DE' 0.028944969177246094
INFO:pyaf.std:START_FORECASTING 'GB'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'GB' 0.029288053512573242
INFO:pyaf.std:START_FORECASTING 'US'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'US' 0.030891895294189453
INFO:pyaf.std:START_FORECASTING 'AMERICA'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'AMERICA' 0.030309677124023438
INFO:pyaf.std:START_FORECASTING 'ASIA'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ASIA' 0.03577613830566406
INFO:pyaf.std:START_FORECASTING 'EUROPE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'EUROPE' 0.027979612350463867
INFO:pyaf.std:START_FORECASTING 'WORLD'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'WORLD' 0.03704357147216797
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODEL_COMBINATION_METHODS ['BU']
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODEL_BOTTOM_UP_METHOD BU
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODEL_OPTIMAL_COMBINATION_METHOD
INFO:pyaf.hierarchical:STRUCTURE [0, 1, 2, 3]
INFO:pyaf.hierarchical:DATASET_COLUMNS Index(['Month', 'ALSACE_BLANC_BE', 'ALSACE_BLANC_BE_Forecast',
       'ALSACE_BLANC_BE_Forecast_Lower_Bound',
       'ALSACE_BLANC_BE_Forecast_Upper_Bound', 'ALSACE_BLANC_CN',
       'ALSACE_BLANC_CN_Forecast', 'ALSACE_BLANC_CN_Forecast_Lower_Bound',
       'ALSACE_BLANC_CN_Forecast_Upper_Bound', 'ALSACE_BLANC_DE',
       ...
       'BORDEAUX_ROUGE_US_BU_Forecast', 'BE_BU_Forecast', 'CN_BU_Forecast',
       'DE_BU_Forecast', 'GB_BU_Forecast', 'US_BU_Forecast',
       'AMERICA_BU_Forecast', 'ASIA_BU_Forecast', 'EUROPE_BU_Forecast',
       'WORLD_BU_Forecast'],
      dtype='object', length=146)
INFO:pyaf.hierarchical:STRUCTURE_LEVEL (0, ['ALSACE_BLANC_BE', 'ALSACE_BLANC_CN', 'ALSACE_BLANC_DE', 'ALSACE_BLANC_GB', 'ALSACE_BLANC_US', 'BEAUJOLAIS_ROUGE_BE', 'BEAUJOLAIS_ROUGE_CN', 'BEAUJOLAIS_ROUGE_DE', 'BEAUJOLAIS_ROUGE_GB', 'BEAUJOLAIS_ROUGE_US', 'BORDEAUX_BLANC_BE', 'BORDEAUX_BLANC_CN', 'BORDEAUX_BLANC_DE', 'BORDEAUX_BLANC_GB', 'BORDEAUX_BLANC_US', 'BORDEAUX_ROUGE_BE', 'BORDEAUX_ROUGE_CN', 'BORDEAUX_ROUGE_DE', 'BORDEAUX_ROUGE_GB', 'BORDEAUX_ROUGE_US'])
INFO:pyaf.hierarchical:MODEL_LEVEL (0, ['ALSACE_BLANC_BE', 'ALSACE_BLANC_CN', 'ALSACE_BLANC_DE', 'ALSACE_BLANC_GB', 'ALSACE_BLANC_US', 'BEAUJOLAIS_ROUGE_BE', 'BEAUJOLAIS_ROUGE_CN', 'BEAUJOLAIS_ROUGE_DE', 'BEAUJOLAIS_ROUGE_GB', 'BEAUJOLAIS_ROUGE_US', 'BORDEAUX_BLANC_BE', 'BORDEAUX_BLANC_CN', 'BORDEAUX_BLANC_DE', 'BORDEAUX_BLANC_GB', 'BORDEAUX_BLANC_US', 'BORDEAUX_ROUGE_BE', 'BORDEAUX_ROUGE_CN', 'BORDEAUX_ROUGE_DE', 'BORDEAUX_ROUGE_GB', 'BORDEAUX_ROUGE_US'])
INFO:pyaf.hierarchical:STRUCTURE_LEVEL (1, ['BE', 'CN', 'DE', 'GB', 'US'])
INFO:pyaf.hierarchical:MODEL_LEVEL (1, ['BE', 'CN', 'DE', 'GB', 'US'])
INFO:pyaf.hierarchical:STRUCTURE_LEVEL (2, ['AMERICA', 'ASIA', 'EUROPE'])
INFO:pyaf.hierarchical:MODEL_LEVEL (2, ['AMERICA', 'ASIA', 'EUROPE'])
INFO:pyaf.hierarchical:STRUCTURE_LEVEL (3, ['WORLD'])
INFO:pyaf.hierarchical:MODEL_LEVEL (3, ['WORLD'])
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'ALSACE_BLANC_BE_BU_Forecast', 'Length': 36, 'MAPE': 0.1418, 'RMSE': 199577.56080021942, 'MAE': 165591.68055555556, 'SMAPE': 0.1379, 'ErrorMean': 0.0, 'ErrorStdDev': 199577.56080021942, 'R2': 0.0, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'ALSACE_BLANC_BE_BU_Forecast', 'Length': 9, 'MAPE': 0.2979, 'RMSE': 444837.29072987556, 'MAE': 326942.97222222225, 'SMAPE': 0.2631, 'ErrorMean': -79348.75, 'ErrorStdDev': 437703.08554696455, 'R2': -0.032864040938387795, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'ALSACE_BLANC_CN_BU_Forecast', 'Length': 36, 'MAPE': 0.4769, 'RMSE': 54489.48176893604, 'MAE': 43502.069322031304, 'SMAPE': 0.3645, 'ErrorMean': -6.063298011819521e-13, 'ErrorStdDev': 54489.48176893604, 'R2': 0.20977709881983608, 'Pearson': 0.4592342911941588}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'ALSACE_BLANC_CN_BU_Forecast', 'Length': 9, 'MAPE': 0.3401, 'RMSE': 86153.69023954055, 'MAE': 54624.102041553844, 'SMAPE': 0.4141, 'ErrorMean': -40494.168775788356, 'ErrorStdDev': 76043.938858062, 'R2': -0.08384682031687452, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'ALSACE_BLANC_DE_BU_Forecast', 'Length': 36, 'MAPE': 0.3037, 'RMSE': 381643.9708516124, 'MAE': 239057.80555555556, 'SMAPE': 0.3884, 'ErrorMean': -235745.16666666666, 'ErrorStdDev': 300127.2011675914, 'R2': -0.5439845652964985, 'Pearson': 0.4222538037199401}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'ALSACE_BLANC_DE_BU_Forecast', 'Length': 9, 'MAPE': 0.1906, 'RMSE': 144769.4922736975, 'MAE': 112843.61111111117, 'SMAPE': 0.2201, 'ErrorMean': -112843.61111111117, 'ErrorStdDev': 90689.16872807068, 'R2': -1.6785920899742903, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'ALSACE_BLANC_GB_BU_Forecast', 'Length': 36, 'MAPE': 0.2964, 'RMSE': 127746.35430699537, 'MAE': 106249.85185185184, 'SMAPE': 0.264, 'ErrorMean': -2.5870071517096624e-11, 'ErrorStdDev': 127746.35430699537, 'R2': 0.0, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'ALSACE_BLANC_GB_BU_Forecast', 'Length': 9, 'MAPE': 0.1402, 'RMSE': 66480.16663925201, 'MAE': 55017.46913580247, 'SMAPE': 0.1358, 'ErrorMean': -731.2222222222481, 'ErrorStdDev': 66476.1451232278, 'R2': -0.00012099491204886625, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'ALSACE_BLANC_US_BU_Forecast', 'Length': 36, 'MAPE': 0.2222, 'RMSE': 165735.85694123103, 'MAE': 131398.81721645, 'SMAPE': 0.1986, 'ErrorMean': -6.467517879274156e-12, 'ErrorStdDev': 165735.85694123103, 'R2': 0.25091278856867993, 'Pearson': 0.5009119569032863}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'ALSACE_BLANC_US_BU_Forecast', 'Length': 9, 'MAPE': 0.238, 'RMSE': 178019.5721837483, 'MAE': 142116.45550728482, 'SMAPE': 0.21, 'ErrorMean': 27404.560804801127, 'ErrorStdDev': 175897.57851539835, 'R2': -0.10452085387212473, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'AMERICA_BU_Forecast', 'Length': 36, 'MAPE': 0.1917, 'RMSE': 2510501.287288395, 'MAE': 1922713.7690322092, 'SMAPE': 0.1847, 'ErrorMean': -8.27842288547092e-10, 'ErrorStdDev': 2510501.287288395, 'R2': 0.037907811620797593, 'Pearson': 0.20054881525310836}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'AMERICA_BU_Forecast', 'Length': 9, 'MAPE': 0.2209, 'RMSE': 2096609.4319779796, 'MAE': 1789928.428711845, 'SMAPE': 0.2023, 'ErrorMean': 47699.14401481321, 'ErrorStdDev': 2096066.7694325198, 'R2': -0.04914130046529963, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'ASIA_BU_Forecast', 'Length': 36, 'MAPE': 0.3787, 'RMSE': 5843680.673765567, 'MAE': 4760283.971866047, 'SMAPE': 0.4745, 'ErrorMean': -3220956.760186489, 'ErrorStdDev': 4875863.140609071, 'R2': -2.1808950571472274, 'Pearson': -0.18079022667756428}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'ASIA_BU_Forecast', 'Length': 9, 'MAPE': 0.2353, 'RMSE': 2852608.7810445037, 'MAE': 2276336.9642346404, 'SMAPE': 0.1973, 'ErrorMean': -122512.68660265993, 'ErrorStdDev': 2849976.7541707437, 'R2': 0.19246072545988602, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BEAUJOLAIS_ROUGE_BE_BU_Forecast', 'Length': 36, 'MAPE': 0.5435, 'RMSE': 106802.3837449867, 'MAE': 80457.65319272065, 'SMAPE': 0.4144, 'ErrorMean': 15233.682772240725, 'ErrorStdDev': 105710.37831171608, 'R2': -0.024414222050409773, 'Pearson': -0.004087940353046451}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BEAUJOLAIS_ROUGE_BE_BU_Forecast', 'Length': 9, 'MAPE': 0.4458, 'RMSE': 126806.9861784883, 'MAE': 82815.0405050574, 'SMAPE': 0.6497, 'ErrorMean': -67684.41089172459, 'ErrorStdDev': 107232.60822115405, 'R2': -0.3166836761041538, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BEAUJOLAIS_ROUGE_CN_BU_Forecast', 'Length': 36, 'MAPE': 1.5835, 'RMSE': 386578.75642473, 'MAE': 186893.800154321, 'SMAPE': 1.9954, 'ErrorMean': -70248.17206790124, 'ErrorStdDev': 380142.51175053976, 'R2': -13.912792387418225, 'Pearson': -0.060453437527646335}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BEAUJOLAIS_ROUGE_CN_BU_Forecast', 'Length': 9, 'MAPE': 1.0, 'RMSE': 190554.48176906724, 'MAE': 147575.0, 'SMAPE': 2.0, 'ErrorMean': -147575.0, 'ErrorStdDev': 120551.3579238234, 'R2': -1.498584962180829, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BEAUJOLAIS_ROUGE_DE_BU_Forecast', 'Length': 36, 'MAPE': 0.6182, 'RMSE': 268908.8411572105, 'MAE': 121252.86111111111, 'SMAPE': 0.5239, 'ErrorMean': -90871.33333333333, 'ErrorStdDev': 253089.63951678487, 'R2': -0.1235609513710596, 'Pearson': 0.17563597266024822}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BEAUJOLAIS_ROUGE_DE_BU_Forecast', 'Length': 9, 'MAPE': 0.6144, 'RMSE': 155871.73582628762, 'MAE': 101230.22222222222, 'SMAPE': 0.6012, 'ErrorMean': -55130.88888888889, 'ErrorStdDev': 145796.375537326, 'R2': -0.1492885450742618, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BEAUJOLAIS_ROUGE_GB_BU_Forecast', 'Length': 36, 'MAPE': 0.7216, 'RMSE': 574972.8645601943, 'MAE': 422320.22222222225, 'SMAPE': 0.5076, 'ErrorMean': 1963.388888888889, 'ErrorStdDev': 574969.5123088063, 'R2': -0.6540904308136148, 'Pearson': 0.16869668407790478}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BEAUJOLAIS_ROUGE_GB_BU_Forecast', 'Length': 9, 'MAPE': 0.3584, 'RMSE': 485564.32192152576, 'MAE': 335481.1111111111, 'SMAPE': 0.3897, 'ErrorMean': -82966.66666666667, 'ErrorStdDev': 478423.7065043217, 'R2': -0.1714086691299752, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BEAUJOLAIS_ROUGE_US_BU_Forecast', 'Length': 36, 'MAPE': 0.5762, 'RMSE': 867034.3960889542, 'MAE': 582255.0717515774, 'SMAPE': 0.4781, 'ErrorMean': 1.8109050061967638e-10, 'ErrorStdDev': 867034.3960889542, 'R2': 0.0028471423817205688, 'Pearson': 0.05534550225360017}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BEAUJOLAIS_ROUGE_US_BU_Forecast', 'Length': 9, 'MAPE': 0.2867, 'RMSE': 774435.6013471647, 'MAE': 387408.1202543912, 'SMAPE': 0.2956, 'ErrorMean': -153564.98378074175, 'ErrorStdDev': 759057.5053251005, 'R2': -0.02749876936315787, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BE_BU_Forecast', 'Length': 36, 'MAPE': 0.1263, 'RMSE': 1158430.1818586537, 'MAE': 870585.8539859032, 'SMAPE': 0.1235, 'ErrorMean': 15233.682772240834, 'ErrorStdDev': 1158330.013921019, 'R2': 0.08023471545578531, 'Pearson': 0.2838047668764661}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BE_BU_Forecast', 'Length': 9, 'MAPE': 0.1993, 'RMSE': 1355866.3595957132, 'MAE': 1199399.0626261914, 'SMAPE': 0.2043, 'ErrorMean': -389555.3695622764, 'ErrorStdDev': 1298699.4260137833, 'R2': 0.10495568930123467, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BORDEAUX_BLANC_BE_BU_Forecast', 'Length': 36, 'MAPE': 0.1248, 'RMSE': 71511.23866079607, 'MAE': 57284.517524619594, 'SMAPE': 0.1211, 'ErrorMean': -1.616879469818539e-12, 'ErrorStdDev': 71511.23866079607, 'R2': 0.29465347365490646, 'Pearson': 0.5433689983428429}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BORDEAUX_BLANC_BE_BU_Forecast', 'Length': 9, 'MAPE': 0.3373, 'RMSE': 133791.73715896055, 'MAE': 125843.55513525108, 'SMAPE': 0.3039, 'ErrorMean': 28407.455518913033, 'ErrorStdDev': 130741.13890797098, 'R2': 0.24879767521606466, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BORDEAUX_BLANC_CN_BU_Forecast', 'Length': 36, 'MAPE': 0.4861, 'RMSE': 236751.57395505198, 'MAE': 196901.17592592593, 'SMAPE': 0.4008, 'ErrorMean': -2.5870071517096624e-11, 'ErrorStdDev': 236751.57395505198, 'R2': 0.0, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BORDEAUX_BLANC_CN_BU_Forecast', 'Length': 9, 'MAPE': 0.306, 'RMSE': 477151.95433478977, 'MAE': 234114.8703703704, 'SMAPE': 0.3129, 'ErrorMean': -116613.27777777782, 'ErrorStdDev': 462682.7541322803, 'R2': -0.06352276318980388, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BORDEAUX_BLANC_DE_BU_Forecast', 'Length': 36, 'MAPE': 0.3966, 'RMSE': 270676.3073877788, 'MAE': 178217.47222222222, 'SMAPE': 0.3632, 'ErrorMean': -18718.194444444445, 'ErrorStdDev': 270028.31810353394, 'R2': -0.18936319314386085, 'Pearson': 0.3559880158188731}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BORDEAUX_BLANC_DE_BU_Forecast', 'Length': 9, 'MAPE': 0.2533, 'RMSE': 201387.2590557804, 'MAE': 97102.88888888889, 'SMAPE': 0.1866, 'ErrorMean': 76439.11111111111, 'ErrorStdDev': 186316.64016545383, 'R2': -13.060718941617052, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BORDEAUX_BLANC_GB_BU_Forecast', 'Length': 36, 'MAPE': 0.259, 'RMSE': 282372.93395753554, 'MAE': 228564.85401399242, 'SMAPE': 0.2421, 'ErrorMean': 3.233758939637078e-12, 'ErrorStdDev': 282372.93395753554, 'R2': 0.14947145177295817, 'Pearson': 0.5128583610185686}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BORDEAUX_BLANC_GB_BU_Forecast', 'Length': 9, 'MAPE': 0.133, 'RMSE': 99936.68789248545, 'MAE': 84699.5553837494, 'SMAPE': 0.1238, 'ErrorMean': 44758.01780843718, 'ErrorStdDev': 89353.57535532457, 'R2': -0.19706891807788685, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BORDEAUX_BLANC_US_BU_Forecast', 'Length': 36, 'MAPE': 0.383, 'RMSE': 403395.3115675406, 'MAE': 307343.7377127167, 'SMAPE': 0.3217, 'ErrorMean': -1.358178754647573e-10, 'ErrorStdDev': 403395.3115675405, 'R2': 0.041980024342557765, 'Pearson': 0.21104598132953822}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BORDEAUX_BLANC_US_BU_Forecast', 'Length': 9, 'MAPE': 0.2096, 'RMSE': 252977.83561553687, 'MAE': 167581.60529117833, 'SMAPE': 0.2035, 'ErrorMean': -54277.80173525347, 'ErrorStdDev': 247086.43336191104, 'R2': -0.21729575877784302, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BORDEAUX_ROUGE_BE_BU_Forecast', 'Length': 36, 'MAPE': 0.1511, 'RMSE': 1034134.42595624, 'MAE': 745887.7208675914, 'SMAPE': 0.1449, 'ErrorMean': 1.2935035758548313e-10, 'ErrorStdDev': 1034134.42595624, 'R2': 0.1148498112118489, 'Pearson': 0.34110079343525745}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BORDEAUX_ROUGE_BE_BU_Forecast', 'Length': 9, 'MAPE': 0.1915, 'RMSE': 921082.2727095841, 'MAE': 803425.001323376, 'SMAPE': 0.1977, 'ErrorMean': -270929.66418946494, 'ErrorStdDev': 880334.9761096833, 'R2': 0.053418461566188125, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BORDEAUX_ROUGE_CN_BU_Forecast', 'Length': 36, 'MAPE': 0.3932, 'RMSE': 5899859.7343801735, 'MAE': 4710841.45749947, 'SMAPE': 0.521, 'ErrorMean': -3569399.141589253, 'ErrorStdDev': 4697630.748939554, 'R2': -2.539579776876861, 'Pearson': -0.19228543926286962}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BORDEAUX_ROUGE_CN_BU_Forecast', 'Length': 9, 'MAPE': 0.2216, 'RMSE': 2646130.4232281283, 'MAE': 2110502.4689600933, 'SMAPE': 0.1932, 'ErrorMean': -382238.5782750007, 'ErrorStdDev': 2618377.338355146, 'R2': 0.2772267547872993, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BORDEAUX_ROUGE_DE_BU_Forecast', 'Length': 36, 'MAPE': 0.337, 'RMSE': 2172921.2151209507, 'MAE': 1699796.3050877494, 'SMAPE': 0.2917, 'ErrorMean': -91240.87311386141, 'ErrorStdDev': 2171004.7697313167, 'R2': -0.046690116723479, 'Pearson': -0.29888213817842174}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BORDEAUX_ROUGE_DE_BU_Forecast', 'Length': 9, 'MAPE': 0.272, 'RMSE': 1186136.6973808932, 'MAE': 910725.4387466401, 'SMAPE': 0.2118, 'ErrorMean': 677534.4440548599, 'ErrorStdDev': 973584.7893187959, 'R2': -0.2559907176181191, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BORDEAUX_ROUGE_GB_BU_Forecast', 'Length': 36, 'MAPE': 0.3547, 'RMSE': 4522958.716730319, 'MAE': 3622128.5, 'SMAPE': 0.3104, 'ErrorMean': 275448.1111111111, 'ErrorStdDev': 4514563.532760625, 'R2': 0.6230681985385662, 'Pearson': 0.8098760167727042}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BORDEAUX_ROUGE_GB_BU_Forecast', 'Length': 9, 'MAPE': 0.1592, 'RMSE': 1259434.9100598427, 'MAE': 951526.7777777778, 'SMAPE': 0.1541, 'ErrorMean': -18088.777777777777, 'ErrorStdDev': 1259305.0022913236, 'R2': -0.6548872013161662, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'BORDEAUX_ROUGE_US_BU_Forecast', 'Length': 36, 'MAPE': 0.3846, 'RMSE': 3670201.8525633416, 'MAE': 3004362.8880813215, 'SMAPE': 0.4796, 'ErrorMean': -1559385.3195952456, 'ErrorStdDev': 3322453.7714752657, 'R2': -3.6041850530076776, 'Pearson': -0.2210489590109037}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'BORDEAUX_ROUGE_US_BU_Forecast', 'Length': 9, 'MAPE': 0.1988, 'RMSE': 1279973.40760195, 'MAE': 1187013.801149747, 'SMAPE': 0.1863, 'ErrorMean': 88484.9492303933, 'ErrorStdDev': 1276911.249041155, 'R2': 0.1688012406765682, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'CN_BU_Forecast', 'Length': 36, 'MAPE': 0.3828, 'RMSE': 6060931.46109166, 'MAE': 4880893.700146054, 'SMAPE': 0.4977, 'ErrorMean': -3639647.3136571553, 'ErrorStdDev': 4846427.303513274, 'R2': -2.4218040424796645, 'Pearson': -0.19318225959972254}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'CN_BU_Forecast', 'Length': 9, 'MAPE': 0.2306, 'RMSE': 2851654.6793566504, 'MAE': 2420007.4080430586, 'SMAPE': 0.2083, 'ErrorMean': -686921.024828567, 'ErrorStdDev': 2767683.8540456803, 'R2': 0.1930008245727045, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'DE_BU_Forecast', 'Length': 36, 'MAPE': 0.2925, 'RMSE': 2544167.6664504567, 'MAE': 1934557.116816145, 'SMAPE': 0.2714, 'ErrorMean': -436575.5675583059, 'ErrorStdDev': 2506429.909018623, 'R2': -0.000994242046815419, 'Pearson': 0.21952310702899056}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'DE_BU_Forecast', 'Length': 9, 'MAPE': 0.2071, 'RMSE': 1175225.5046689187, 'MAE': 933131.1733145413, 'SMAPE': 0.1745, 'ErrorMean': 585999.0551659709, 'ErrorStdDev': 1018705.1065783974, 'R2': -0.11602593999046817, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'EUROPE_BU_Forecast', 'Length': 36, 'MAPE': 0.212, 'RMSE': 6747586.225448739, 'MAE': 5705896.63604811, 'SMAPE': 0.2008, 'ErrorMean': -159164.06755830607, 'ErrorStdDev': 6745708.759608871, 'R2': 0.3741000559540115, 'Pearson': 0.6477186996501847}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'EUROPE_BU_Forecast', 'Length': 9, 'MAPE': 0.1116, 'RMSE': 2443759.540347075, 'MAE': 2059543.2845683566, 'SMAPE': 0.1077, 'ErrorMean': 278069.3076139353, 'ErrorStdDev': 2427887.590313945, 'R2': 0.4954231535971201, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'GB_BU_Forecast', 'Length': 36, 'MAPE': 0.3427, 'RMSE': 4910323.308976409, 'MAE': 4088050.915260217, 'SMAPE': 0.2992, 'ErrorMean': 277411.5000000002, 'ErrorStdDev': 4902480.786127038, 'R2': 0.5807302960568244, 'Pearson': 0.7856609595821422}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'GB_BU_Forecast', 'Length': 9, 'MAPE': 0.1514, 'RMSE': 1555255.9635870806, 'MAE': 1179264.328081761, 'SMAPE': 0.1455, 'ErrorMean': -57028.64885822953, 'ErrorStdDev': 1554210.0390496075, 'R2': -0.32675474980832786, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'US_BU_Forecast', 'Length': 36, 'MAPE': 0.3174, 'RMSE': 4139444.8981735716, 'MAE': 3423881.523455315, 'SMAPE': 0.363, 'ErrorMean': -1559385.3195952454, 'ErrorStdDev': 3834491.034028643, 'R2': -1.615652245446988, 'Pearson': -0.1851354794911439}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'US_BU_Forecast', 'Length': 9, 'MAPE': 0.2064, 'RMSE': 1961288.3969814766, 'MAE': 1729787.2463276673, 'SMAPE': 0.1953, 'ErrorMean': -91953.27548080104, 'ErrorStdDev': 1959131.6370429327, 'R2': 0.08191724153051494, 'Pearson': 0.0}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_FIT_PERF {'Signal': 'WORLD_BU_Forecast', 'Length': 36, 'MAPE': 0.147, 'RMSE': 9271518.686280593, 'MAE': 7579789.866852175, 'SMAPE': 0.1531, 'ErrorMean': -3380120.8277447955, 'ErrorStdDev': 8633414.268983973, 'R2': 0.31923877147646307, 'Pearson': 0.6412775585974727}
INFO:pyaf.hierarchical:REPORT_COMBINED_FORECASTS_VALID_PERF {'Signal': 'WORLD_BU_Forecast', 'Length': 9, 'MAPE': 0.1213, 'RMSE': 5720232.5387728, 'MAE': 4423326.626083633, 'SMAPE': 0.1129, 'ErrorMean': 203255.76502609006, 'ErrorStdDev': 5716620.2770184735, 'R2': 0.3573359608539569, 'Pearson': 0.0}
INFO:pyaf.std:END_HIERARCHICAL_TRAINING_TIME_IN_SECONDS 16.736175298690796

In [20]:

    

French_Wine_Export_in_Euros_DF.info()


    

    




<class 'pandas.core.frame.DataFrame'>
RangeIndex: 50 entries, 0 to 49
Columns: 177 entries, Month to RHÔNE_ROUGE_US
dtypes: datetime64[ns](1), float64(56), int64(120)
memory usage: 69.3 KB

In [21]:

    

lInfo = lEngine.to_json()
lInfo.keys()


    

    
Out[21]:





dict_keys(['Structure', 'Models'])

In [22]:

    

lInfo['Structure']


    

    
Out[22]:





{0: {'ALSACE_BLANC_GB': set(),
  'ALSACE_BLANC_US': set(),
  'ALSACE_BLANC_DE': set(),
  'ALSACE_BLANC_BE': set(),
  'ALSACE_BLANC_CN': set(),
  'BORDEAUX_BLANC_GB': set(),
  'BORDEAUX_BLANC_US': set(),
  'BORDEAUX_BLANC_DE': set(),
  'BORDEAUX_BLANC_BE': set(),
  'BORDEAUX_BLANC_CN': set(),
  'BEAUJOLAIS_ROUGE_GB': set(),
  'BEAUJOLAIS_ROUGE_US': set(),
  'BEAUJOLAIS_ROUGE_DE': set(),
  'BEAUJOLAIS_ROUGE_BE': set(),
  'BEAUJOLAIS_ROUGE_CN': set(),
  'BORDEAUX_ROUGE_GB': set(),
  'BORDEAUX_ROUGE_US': set(),
  'BORDEAUX_ROUGE_DE': set(),
  'BORDEAUX_ROUGE_BE': set(),
  'BORDEAUX_ROUGE_CN': set()},
 1: {'GB': {'ALSACE_BLANC_GB',
   'BEAUJOLAIS_ROUGE_GB',
   'BORDEAUX_BLANC_GB',
   'BORDEAUX_ROUGE_GB'},
  'US': {'ALSACE_BLANC_US',
   'BEAUJOLAIS_ROUGE_US',
   'BORDEAUX_BLANC_US',
   'BORDEAUX_ROUGE_US'},
  'DE': {'ALSACE_BLANC_DE',
   'BEAUJOLAIS_ROUGE_DE',
   'BORDEAUX_BLANC_DE',
   'BORDEAUX_ROUGE_DE'},
  'BE': {'ALSACE_BLANC_BE',
   'BEAUJOLAIS_ROUGE_BE',
   'BORDEAUX_BLANC_BE',
   'BORDEAUX_ROUGE_BE'},
  'CN': {'ALSACE_BLANC_CN',
   'BEAUJOLAIS_ROUGE_CN',
   'BORDEAUX_BLANC_CN',
   'BORDEAUX_ROUGE_CN'}},
 2: {'EUROPE': {'BE', 'DE', 'GB'}, 'AMERICA': {'US'}, 'ASIA': {'CN'}},
 3: {'WORLD': {'AMERICA', 'ASIA', 'EUROPE'}}}

In [23]:

    

lInfo['Models'].keys()


    

    
Out[23]:





dict_keys(['ALSACE_BLANC_BE', 'ALSACE_BLANC_CN', 'ALSACE_BLANC_DE', 'ALSACE_BLANC_GB', 'ALSACE_BLANC_US', 'BEAUJOLAIS_ROUGE_BE', 'BEAUJOLAIS_ROUGE_CN', 'BEAUJOLAIS_ROUGE_DE', 'BEAUJOLAIS_ROUGE_GB', 'BEAUJOLAIS_ROUGE_US', 'BORDEAUX_BLANC_BE', 'BORDEAUX_BLANC_CN', 'BORDEAUX_BLANC_DE', 'BORDEAUX_BLANC_GB', 'BORDEAUX_BLANC_US', 'BORDEAUX_ROUGE_BE', 'BORDEAUX_ROUGE_CN', 'BORDEAUX_ROUGE_DE', 'BORDEAUX_ROUGE_GB', 'BORDEAUX_ROUGE_US', 'BE', 'CN', 'DE', 'GB', 'US', 'AMERICA', 'ASIA', 'EUROPE', 'WORLD'])

In [24]:

    

lInfo['Models']['BORDEAUX_ROUGE_CN']


    

    
Out[24]:





{'Dataset': {'Time': {'TimeVariable': 'Month',
   'TimeMinMax': ['2012-01-01 00:00:00', '2016-05-01 00:00:00'],
   'Horizon': 1},
  'Signal': 'BORDEAUX_ROUGE_CN',
  'Training_Signal_Length': 46},
 'Model': {'Best_Decomposition': 'Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_AR(11)',
  'Signal_Transoformation': 'Difference',
  'Trend': 'ConstantTrend',
  'Cycle': 'NoCycle',
  'AR_Model': 'AR'},
 'Model_Performance': {'MAPE': '0.2216',
  'MASE': '0.7022',
  'MAE': '2110502.4689600933',
  'RMSE': '2646130.4232281283',
  'COMPLEXITY': '41'}}

In [25]:

    

perfs = [];
for model in sorted(lInfo['Models'].keys()):
    lPerf = lInfo['Models'][model]['Model_Performance']
    perfs.append([model , lPerf['RMSE'] , lPerf['MAPE']])
df_perf = pd.DataFrame(perfs , columns=['Model' , 'RMSE' , 'MAPE']);
df_perf = df_perf.sort_values(by = ['MAPE'])
print(df_perf)


    

    




                  Model                RMSE    MAPE
16    BORDEAUX_BLANC_GB   99936.68789248545   0.133
28                WORLD   6516614.471320467  0.1367
3       ALSACE_BLANC_GB   66480.16663925201  0.1402
25               EUROPE   3085512.630860862  0.1468
26                   GB  1567222.4394758958  0.1579
21    BORDEAUX_ROUGE_GB  1259434.9100598427  0.1592
7                    BE  1150941.4629769188  0.1693
2       ALSACE_BLANC_DE   144769.4922736975  0.1906
18    BORDEAUX_ROUGE_BE   921082.2727095841  0.1915
22    BORDEAUX_ROUGE_US    1279973.40760195  0.1988
17    BORDEAUX_BLANC_US  252977.83561553687  0.2096
5               AMERICA  2096609.4319779796  0.2209
27                   US  2096609.4319779796  0.2209
19    BORDEAUX_ROUGE_CN  2646130.4232281283  0.2216
23                   CN  2852608.7810445037  0.2353
6                  ASIA  2852608.7810445037  0.2353
24                   DE  1311702.0584794695  0.2377
4       ALSACE_BLANC_US   178019.5721837483   0.238
15    BORDEAUX_BLANC_DE   201387.2590557804  0.2533
20    BORDEAUX_ROUGE_DE  1186136.6973808932   0.272
12  BEAUJOLAIS_ROUGE_US   774435.6013471647  0.2867
0       ALSACE_BLANC_BE  444837.29072987556  0.2979
14    BORDEAUX_BLANC_CN  477151.95433478977   0.306
13    BORDEAUX_BLANC_BE  133791.73715896055  0.3373
1       ALSACE_BLANC_CN   86153.69023954055  0.3401
11  BEAUJOLAIS_ROUGE_GB  485564.32192152576  0.3584
8   BEAUJOLAIS_ROUGE_BE   126806.9861784883  0.4458
10  BEAUJOLAIS_ROUGE_DE  155871.73582628762  0.6144
9   BEAUJOLAIS_ROUGE_CN  190554.48176906724     1.0

In [26]:

    

lEngine.mSignalHierarchy.plot()


    

    




INFO:pyaf.std:START_HIERARCHICAL_PLOTTING






    













    




INFO:pyaf.std:END_HIERARCHICAL_PLOTTING_TIME_IN_SECONDS 0.25464677810668945

In [27]:

    

BORDEAUX_ROUGE_CN_Engine = lEngine.mSignalHierarchy.mModels[0]['BORDEAUX_ROUGE_CN']


    

In [28]:

    

BORDEAUX_ROUGE_CN_Engine.getModelInfo()


    

    




INFO:pyaf.std:TIME_DETAIL TimeVariable='Month' TimeMin=2012-01-01T00:00:00.000000 TimeMax=2014-12-01T00:00:00.000000 TimeDelta=<DateOffset: months=1> Horizon=1
INFO:pyaf.std:SIGNAL_DETAIL_ORIG SignalVariable='BORDEAUX_ROUGE_CN' Length=46  Min=5377235 Max=19050876  Mean=11306225.717391305 StdDev=3097054.2409197744
INFO:pyaf.std:SIGNAL_DETAIL_TRANSFORMED TransformedSignalVariable='Diff_BORDEAUX_ROUGE_CN' Min=-7118004.0 Max=8508070.0  Mean=58285.34782608696 StdDev=2959960.353380101
INFO:pyaf.std:BEST_TRANSOFORMATION_TYPE 'Diff_'
INFO:pyaf.std:BEST_DECOMPOSITION  'Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_AR(11)' [ConstantTrend + NoCycle + AR]
INFO:pyaf.std:TREND_DETAIL 'Diff_BORDEAUX_ROUGE_CN_ConstantTrend' [ConstantTrend]
INFO:pyaf.std:CYCLE_DETAIL 'Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle' [NoCycle]
INFO:pyaf.std:AUTOREG_DETAIL 'Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_AR(11)' [AR]
INFO:pyaf.std:MODEL_MAPE MAPE_Fit=0.3932 MAPE_Forecast=0.2216 MAPE_Test=0.0703
INFO:pyaf.std:MODEL_SMAPE SMAPE_Fit=0.521 SMAPE_Forecast=0.1932 SMAPE_Test=0.0679
INFO:pyaf.std:MODEL_MASE MASE_Fit=2.2581 MASE_Forecast=0.7022 MASE_Test=None
INFO:pyaf.std:MODEL_L1 L1_Fit=4710841.45749947 L1_Forecast=2110502.4689600933 L1_Test=793876.8093173634
INFO:pyaf.std:MODEL_L2 L2_Fit=5899859.7343801735 L2_Forecast=2646130.4232281283 L2_Test=793876.8093173634
INFO:pyaf.std:MODEL_COMPLEXITY 41
INFO:pyaf.std:SIGNAL_TRANSFORMATION_DETAIL_START
INFO:pyaf.std:DIFFERENCING_TRANSFORMATION Difference 8616750
INFO:pyaf.std:SIGNAL_TRANSFORMATION_DETAIL_END
INFO:pyaf.std:TREND_DETAIL_START
INFO:pyaf.std:CONSTANT_TREND ConstantTrend 102348.25
INFO:pyaf.std:TREND_DETAIL_END
INFO:pyaf.std:CYCLE_MODEL_DETAIL_START
INFO:pyaf.std:ZERO_CYCLE_MODEL_VALUES Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle 0.0 {}
INFO:pyaf.std:CYCLE_MODEL_DETAIL_END
INFO:pyaf.std:AR_MODEL_DETAIL_START
INFO:pyaf.std:AR_MODEL_COEFF 1 Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_Lag3 -0.3867966153192115
INFO:pyaf.std:AR_MODEL_COEFF 2 Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_Lag9 -0.29935738107599374
INFO:pyaf.std:AR_MODEL_COEFF 3 Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_Lag2 -0.2945132457675459
INFO:pyaf.std:AR_MODEL_COEFF 4 Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_Lag6 -0.13283631227578233
INFO:pyaf.std:AR_MODEL_COEFF 5 Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_Lag7 0.10596403673387618
INFO:pyaf.std:AR_MODEL_COEFF 6 Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_Lag10 -0.06307985742933694
INFO:pyaf.std:AR_MODEL_COEFF 7 Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_Lag1 -0.05288492538443518
INFO:pyaf.std:AR_MODEL_COEFF 8 Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_Lag11 -0.03488010234067359
INFO:pyaf.std:AR_MODEL_COEFF 9 Diff_BORDEAUX_ROUGE_CN_ConstantTrend_residue_zeroCycle_residue_Lag5 0.030428207374232505
INFO:pyaf.std:AR_MODEL_DETAIL_END

In [29]:

    

BORDEAUX_ROUGE_CN_Engine.standardPlots()


    

    




INFO:pyaf.std:START_PLOTTING
INFO:pyaf.std:END_PLOTTING_TIME_IN_SECONDS 0.4307279586791992






    













    













    













    













    













    

In [30]:

    

lEngine.mOptions.mHierarchicalCombinationMethod = ["BU" , 'TD' , 'MO' , 'OC'];
dfapp_out = lEngine.forecast(train_df, H);


    

    




INFO:pyaf.std:START_HIERARCHICAL_FORECASTING
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODELS_LEVEL_SIGNAL [(0, 'ALSACE_BLANC_BE'), (0, 'ALSACE_BLANC_CN'), (0, 'ALSACE_BLANC_DE'), (0, 'ALSACE_BLANC_GB'), (0, 'ALSACE_BLANC_US'), (0, 'BEAUJOLAIS_ROUGE_BE'), (0, 'BEAUJOLAIS_ROUGE_CN'), (0, 'BEAUJOLAIS_ROUGE_DE'), (0, 'BEAUJOLAIS_ROUGE_GB'), (0, 'BEAUJOLAIS_ROUGE_US'), (0, 'BORDEAUX_BLANC_BE'), (0, 'BORDEAUX_BLANC_CN'), (0, 'BORDEAUX_BLANC_DE'), (0, 'BORDEAUX_BLANC_GB'), (0, 'BORDEAUX_BLANC_US'), (0, 'BORDEAUX_ROUGE_BE'), (0, 'BORDEAUX_ROUGE_CN'), (0, 'BORDEAUX_ROUGE_DE'), (0, 'BORDEAUX_ROUGE_GB'), (0, 'BORDEAUX_ROUGE_US'), (1, 'BE'), (1, 'CN'), (1, 'DE'), (1, 'GB'), (1, 'US'), (2, 'AMERICA'), (2, 'ASIA'), (2, 'EUROPE'), (3, 'WORLD')]
INFO:pyaf.std:START_FORECASTING 'ALSACE_BLANC_BE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ALSACE_BLANC_BE' 0.10614252090454102
INFO:pyaf.std:START_FORECASTING 'ALSACE_BLANC_CN'
INFO:pyaf.std:START_FORECASTING 'ALSACE_BLANC_DE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ALSACE_BLANC_CN' 0.1340646743774414
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ALSACE_BLANC_DE' 0.09419608116149902
INFO:pyaf.std:START_FORECASTING 'ALSACE_BLANC_GB'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ALSACE_BLANC_GB' 0.09335899353027344
INFO:pyaf.std:START_FORECASTING 'ALSACE_BLANC_US'
INFO:pyaf.std:START_FORECASTING 'BEAUJOLAIS_ROUGE_BE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ALSACE_BLANC_US' 0.1188194751739502
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BEAUJOLAIS_ROUGE_BE' 0.10483503341674805
INFO:pyaf.std:START_FORECASTING 'BEAUJOLAIS_ROUGE_CN'
INFO:pyaf.std:START_FORECASTING 'BEAUJOLAIS_ROUGE_DE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BEAUJOLAIS_ROUGE_CN' 0.09068918228149414
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BEAUJOLAIS_ROUGE_DE' 0.08982563018798828
INFO:pyaf.std:START_FORECASTING 'BEAUJOLAIS_ROUGE_GB'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BEAUJOLAIS_ROUGE_GB' 0.08919095993041992
INFO:pyaf.std:START_FORECASTING 'BEAUJOLAIS_ROUGE_US'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BEAUJOLAIS_ROUGE_US' 0.09915328025817871
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_BLANC_BE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_BLANC_BE' 0.12746596336364746
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_BLANC_CN'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_BLANC_CN' 0.08881258964538574
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_BLANC_DE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_BLANC_DE' 0.1094510555267334
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_BLANC_GB'
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_BLANC_US'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_BLANC_GB' 0.12147212028503418
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_BLANC_US' 0.08834195137023926
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_ROUGE_BE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_ROUGE_BE' 0.09271359443664551
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_ROUGE_CN'
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_ROUGE_DE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_ROUGE_CN' 0.1137382984161377
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_ROUGE_DE' 0.09014034271240234
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_ROUGE_GB'
INFO:pyaf.std:START_FORECASTING 'BORDEAUX_ROUGE_US'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_ROUGE_GB' 0.0887598991394043
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BORDEAUX_ROUGE_US' 0.1569528579711914
INFO:pyaf.std:START_FORECASTING 'BE'
INFO:pyaf.std:START_FORECASTING 'CN'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'BE' 0.126878023147583
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'CN' 0.11184477806091309
INFO:pyaf.std:START_FORECASTING 'DE'
INFO:pyaf.std:START_FORECASTING 'GB'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'DE' 0.07956194877624512
INFO:pyaf.std:START_FORECASTING 'US'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'GB' 0.0834503173828125
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'US' 0.08534836769104004
INFO:pyaf.std:START_FORECASTING 'AMERICA'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'AMERICA' 0.08472228050231934
INFO:pyaf.std:START_FORECASTING 'ASIA'
INFO:pyaf.std:START_FORECASTING 'EUROPE'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'ASIA' 0.10328197479248047
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'EUROPE' 0.08194112777709961
INFO:pyaf.std:START_FORECASTING 'WORLD'
INFO:pyaf.std:END_FORECAST_TIME_IN_SECONDS  'WORLD' 0.12117910385131836
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODEL_COMBINATION_METHODS ['BU', 'TD', 'MO', 'OC']
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODEL_BOTTOM_UP_METHOD BU
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODEL_TOP_DOWN_METHOD AHP_TD
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODEL_TOP_DOWN_METHOD PHA_TD
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODEL_MIDDLE_OUT_METHOD MO
INFO:pyaf.hierarchical:FORECASTING_HIERARCHICAL_MODEL_OPTIMAL_COMBINATION_METHOD OC
INFO:pyaf.std:END_HIERARCHICAL_FORECAST_TIME_IN_SECONDS 4.700767517089844

In [31]:

    

for country in Countries:
    dfapp_out.plot('Month' , 
                    [country , country + '_Forecast' , 
                     country + '_BU_Forecast',  
                     country + '_PHA_TD_Forecast',  
                     country + '_AHP_TD_Forecast'  ,  
                     country + '_MO_Forecast' ,
                     country + '_OC_Forecast'  ],
                figsize=(32 , 12)).legend(fontsize=18)


    

In [32]:

    

for region in ['AMERICA', 'ASIA', 'EUROPE']:
    dfapp_out.plot('Month' , 
                    [region , region + '_Forecast' , 
                     region + '_BU_Forecast',  
                     region + '_PHA_TD_Forecast',  
                     region + '_AHP_TD_Forecast'  ,  
                     region + '_MO_Forecast' ,
                     region + '_OC_Forecast'  ],
                    figsize=(32 , 12)).legend(fontsize=18)


    

In [33]:

    

world = 'WORLD'
dfapp_out.plot('Month' , 
                [world , world + '_Forecast' , 
                 world + '_BU_Forecast',  
                 world + '_PHA_TD_Forecast',  
                 world + '_AHP_TD_Forecast'  ,  
                 world + '_MO_Forecast' ,
                 world + '_OC_Forecast'  ],
                figsize=(32 , 12)).legend(fontsize=18)


    

    
Out[33]:





<matplotlib.legend.Legend at 0x7feb2a516dc0>






    

In [ ]: