

In [1]:

    
import bisect
import pickle

import pandas as pd
import numpy as np

Charge les référentiels

cf referentiels.py



In [2]:

    
ROME_df = pd.read_csv('../referentiels/referentiel_ROME/20150921_arboprincipale28427_ROME.csv', index_col=0, sep='|', dtype=str)
OGR_df = pd.read_csv('../referentiels/referentiel_OGR/20150921_arboprincipale28427_OGR.csv', sep='|', dtype=str).set_index('OGR')
NAF_df = pd.read_csv('../referentiels/referentiel_NAF/naf2008_liste_n5_nouveau_header.csv', sep='|', encoding="utf-8").set_index(['NAF'])



In [3]:

    
tranches_effectif = ['00', '01', '02', '03', '11', '12', '21', '22', '31', '32', '41', '42', '51', '52', '53', 'NN']
seuils_tranches_effectif = [0, 1, 3, 6, 10, 20, 50, 100, 200, 250, 500, 1000, 2000, 5000, 10000]



In [4]:

    
def trouve_tranche_effectif(effectif):
    index = bisect.bisect(seuils_tranches_effectif, effectif) - 1
    return index, tranches_effectif[index]

Charge les DPAE



In [7]:

    
filename = '../LBB_XDPDPA_DPAE_20160302_20170331_20170331_172811_clean.csv'



In [8]:

    
df = pd.read_csv(filename, sep='|', dtype=str, keep_default_na=False)

Array : ROME x NAF x effectif



In [9]:

    
def ponderation_duree(dc_typecontrat_id, nbrjourtravaille):
    if dc_typecontrat_id == '2':  # CDI
        return 1
    elif dc_typecontrat_id == '1':  # CDD
        if int(nbrjourtravaille) >= 30:
            return 1
        else:
            return 0
    raise ValueError()



In [10]:

    
n_NAF = len(NAF_df)
n_ROME = len(ROME_df)
n_tranches = len(tranches_effectif)

array_ROME1 = np.zeros((n_NAF, n_ROME, n_tranches))
array_ROME2 = np.zeros((n_NAF, n_ROME, n_tranches))



In [11]:

    
for i, row in df.iterrows():
    (dc_naf_id, dn_tailleetablissement, kd_dateembauche, dc_typecontrat_id, dd_datefincdd,
     dc_romev3_1_id, dc_romev3_2_id, nbrjourtravaille) = row
    
    if (dc_naf_id in NAF_df.index) and (dc_romev3_1_id in ROME_df.index) and (dc_typecontrat_id in {'1', '2'}):
        
        if dc_typecontrat_id == '1' and nbrjourtravaille == 'NULL':
            continue
        
        NAF_index = NAF_df.index.get_loc(dc_naf_id)
        ROME_index = ROME_df.index.get_loc(dc_romev3_1_id)
        
        try:
            effectif = int(dn_tailleetablissement)
        except ValueError:
            tranche = 'NA'
            tranche_index = 15
        else:
            tranche_index, tranche = trouve_tranche_effectif(effectif)
            
        poids = ponderation_duree(dc_typecontrat_id, nbrjourtravaille)
            
        array_ROME1[NAF_index, ROME_index, tranche_index] += poids

        if dc_romev3_2_id in ROME_df.index:
            ROME2_index = ROME_df.index.get_loc(dc_romev3_2_id)
            array_ROME2[NAF_index, ROME2_index, tranche_index] += poids
        
    if i % 100000 == 0:
        print(i)



In [13]:

    
with open('../array_ROME1_CDI_CDD.pickle', 'wb') as f:
    pickle.dump(array_ROME1, f)



In [14]:

    
with open('../array_ROME2_CDI_CDD.pickle', 'wb') as f:
    pickle.dump(array_ROME2, f)



In [ ]: