In [1]:
%matplotlib inline
import matplotlib as mpl
import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns
from jupyterthemes import jtplot
import re
import unidecode
jtplot.style()
def millions(x, pos):
return '%1.f M€' % (x*1e-6)
def custom_parser(s):
if type(s) == str:
return pd.datetime.strptime(s, '%d/%m/%Y')
return pd.np.nan
In [2]:
avantage_data = pd.read_csv('../../data/tmp/declaration_avantage_2017_07_27_04_00.csv',
parse_dates=['avant_date_signature'],
date_parser=custom_parser,
sep=';')
In [4]:
avantage_data.head()
Out[4]:
In [5]:
print("Total des avantages depuis mi 2013 :", millions(avantage_data.avant_montant_ttc.sum(), ""))
print("Nombre d'avantages :", avantage_data.avant_montant_ttc.shape[0])
print("Montant moyen d'un avantage :", avantage_data.avant_montant_ttc.mean())
In [7]:
avantage_data_by_month = avantage_data.groupby(avantage_data.avant_date_signature.dt.to_period("M"))
avantage_data_by_quarter = avantage_data.groupby(avantage_data.avant_date_signature.dt.to_period("Q"))
avantage_data_by_year = avantage_data.groupby(avantage_data.avant_date_signature.dt.to_period("A"))
avantage_data_by_year.avant_montant_ttc.sum()
Out[7]:
In [8]:
with sns.axes_style("whitegrid"):
sns.set_context("talk")
plot = avantage_data_by_year.avant_montant_ttc.sum().plot()
formatter = mpl.ticker.FuncFormatter(millions)
plot.yaxis.set_major_formatter(formatter)
plot = avantage_data_by_quarter.avant_montant_ttc.sum().plot()
formatter = mpl.ticker.FuncFormatter(millions)
plot.yaxis.set_major_formatter(formatter)
In [9]:
repas_re = re.compile('(repas|restauration|restaurant|dejeuner|diner|collation|bouche|coktail|cocktail|traiteur|buffet|reception)')
transport_re = re.compile('(transport|deplacement|train)')
hebergement_re = re.compile('(hebergement|hospitalite|nuit|voyage|hotel|bergement)')
congre_re = re.compile("(inscription|congres|invitation|conferences)")
autre_re = re.compile("(autre|divers|cadeau)")
don_re = re.compile("don")
def normalize_nature(nature):
nature = unidecode.unidecode(nature).lower()
if repas_re.search(nature):
return "REPAS"
if hebergement_re.search(nature):
return "HEBERGEMENT"
if transport_re.search(nature):
return "TRANSPORT"
if congre_re.search(nature):
return "CONGRES"
if don_re.search(nature):
return "DON"
if autre_re.search(nature):
return "AUTRE"
return nature
avantage_data["normalized_avant_nature"] = avantage_data.avant_nature.apply(normalize_nature)
avant_by_nature = avantage_data.groupby('normalized_avant_nature').sum()
In [10]:
avant_by_nature.reset_index(level=0, inplace=True)
avant_by_nature.sort_values(by="avant_montant_ttc", ascending=False, inplace=True)
In [11]:
with sns.axes_style("whitegrid"):
sns.set_context("talk")
sns.barplot(x="avant_montant_ttc", y="normalized_avant_nature", data=avant_by_nature[:10], label="Total", color="b")
In [ ]: