|
|
@@ -0,0 +1,171 @@
|
|
|
+import math
|
|
|
+import pandas as pd
|
|
|
+from .test import testQualitatif, testQuantitatif
|
|
|
+
|
|
|
+class analyseStatistiques ():
|
|
|
+ """
|
|
|
+ Permet l'analyse d'un jeu de données.
|
|
|
+ Analyses univariés :
|
|
|
+ Description des données
|
|
|
+ Application des tests
|
|
|
+
|
|
|
+ Input : dataset
|
|
|
+ """
|
|
|
+
|
|
|
+ def __init__ (self, df):
|
|
|
+ # Chargement du dataframe
|
|
|
+ self.df = df
|
|
|
+
|
|
|
+ def _describe_qualitative (self, data):
|
|
|
+
|
|
|
+ """
|
|
|
+ Calculate n and p of each modalitie of the qualitative value
|
|
|
+ Input : data, Pandas Series containing data to describe
|
|
|
+ """
|
|
|
+
|
|
|
+ table = data.value_counts()
|
|
|
+ description = pd.DataFrame({'n':table, 'p':table/table.sum()}) \
|
|
|
+ .to_dict("index")
|
|
|
+ description["total"] = table.sum()
|
|
|
+
|
|
|
+ return(description)
|
|
|
+
|
|
|
+ def _get_sub_table(self, variable, axes):
|
|
|
+
|
|
|
+ # On sélectionne les données à analyse
|
|
|
+ if (axes is None):
|
|
|
+ temp_data = self.df[[variable]]
|
|
|
+ else:
|
|
|
+ temp_data = self.df[[variable]+axes]
|
|
|
+
|
|
|
+ temp_data = temp_data.dropna()
|
|
|
+
|
|
|
+ return(temp_data)
|
|
|
+
|
|
|
+
|
|
|
+ def _analyse_univarie_qualitative (self, variable, axes = None):
|
|
|
+
|
|
|
+ temp_data = self._get_sub_table(variable, [])
|
|
|
+
|
|
|
+ # On charge un dictionnaire vide
|
|
|
+ analyse = {}
|
|
|
+
|
|
|
+ analyse["n"] = temp_data.shape[0]
|
|
|
+
|
|
|
+ ## Globale : en dehors de l'axe d'analyse
|
|
|
+ analyse["global"] = self._describe_qualitative(temp_data[variable])
|
|
|
+
|
|
|
+ ## Spécifique : Dans les axes d'analyse
|
|
|
+ if (axes is not None):
|
|
|
+
|
|
|
+ analyse["sous_groupes"] = {}
|
|
|
+ analyse["test"] = {}
|
|
|
+
|
|
|
+ for axe in axes:
|
|
|
+
|
|
|
+ temp_data = self._get_sub_table(variable, [axe])
|
|
|
+
|
|
|
+ # Axe values
|
|
|
+ axe_values = temp_data[axe] \
|
|
|
+ .drop_duplicates().values.tolist()
|
|
|
+
|
|
|
+ # Description
|
|
|
+ analyse["sous_groupes"][axe] = {}
|
|
|
+ for values in axe_values:
|
|
|
+ analyse["sous_groupes"][axe][values] = self._describe_qualitative(
|
|
|
+ temp_data[
|
|
|
+ temp_data[axe] == values
|
|
|
+ ] \
|
|
|
+ .reset_index(drop = True)[variable]
|
|
|
+ )
|
|
|
+
|
|
|
+ # Test statistique
|
|
|
+ analyse["test"][axe] = testQualitatif(temp_data, variable,axe).best_test()
|
|
|
+
|
|
|
+ analyse["type"] = "qualitative"
|
|
|
+
|
|
|
+ return analyse
|
|
|
+
|
|
|
+ def _describe_quantitative (self, data):
|
|
|
+ """
|
|
|
+ Calculate mean, median, Q25, 50, 27, std, std_mean and CI for quantitative data
|
|
|
+ Input : data, Pandas Series containing data to describe
|
|
|
+ """
|
|
|
+
|
|
|
+ # Dict containing data
|
|
|
+ description = {}
|
|
|
+
|
|
|
+ description["n"] = data.shape[0]
|
|
|
+ description["mean"] = data.mean()
|
|
|
+ description["median"] = data.median()
|
|
|
+ description["Q25"] = data.quantile(0.25)
|
|
|
+ description["Q75"] = data.quantile(0.75)
|
|
|
+ description["std"] = data.std()
|
|
|
+ description["std_mean"] = description["std"]/math.sqrt(description["n"])
|
|
|
+ description["ci_95"] = [description["mean"]-1.96*description["std_mean"],
|
|
|
+ description["mean"]+1.96*description["std_mean"]]
|
|
|
+
|
|
|
+ return description
|
|
|
+
|
|
|
+ def _analyse_univarie_quantitative (self, variable, axes = None):
|
|
|
+
|
|
|
+ # On sélectionne les données à analyse
|
|
|
+ temp_data = self._get_sub_table(variable, [])
|
|
|
+
|
|
|
+ # On charge un dictionnaire vide
|
|
|
+ analyse = {}
|
|
|
+
|
|
|
+ analyse["n"] = temp_data.shape[0]
|
|
|
+
|
|
|
+ ## Globale : en dehors de l'axe d'analyse
|
|
|
+ analyse["global"] = self._describe_quantitative(temp_data[variable])
|
|
|
+
|
|
|
+ ## Spécifique : Dans les axes d'analyse
|
|
|
+ if (axes is not None):
|
|
|
+
|
|
|
+ analyse["sous_groupes"] = {}
|
|
|
+ analyse["test"] = {}
|
|
|
+
|
|
|
+ for axe in axes:
|
|
|
+
|
|
|
+ temp_data = self._get_sub_table(variable, [axe])
|
|
|
+
|
|
|
+ # Axe values
|
|
|
+ axe_values = temp_data[axe] \
|
|
|
+ .drop_duplicates().values.tolist()
|
|
|
+
|
|
|
+ # Description
|
|
|
+ analyse["sous_groupes"][axe] = {}
|
|
|
+ for values in axe_values:
|
|
|
+ analyse["sous_groupes"][axe][values] = self._describe_quantitative(
|
|
|
+ temp_data[
|
|
|
+ temp_data[axe] == values
|
|
|
+ ] \
|
|
|
+ .reset_index(drop = True)[variable]
|
|
|
+ )
|
|
|
+
|
|
|
+ # Test statistique
|
|
|
+ analyse["test"][axe] = testQuantitatif(temp_data, variable,axe).best_test()
|
|
|
+
|
|
|
+ analyse["type"] = "quantitative"
|
|
|
+
|
|
|
+ return analyse
|
|
|
+
|
|
|
+ def analyse_univarie (self, variables, axes = None):
|
|
|
+ """
|
|
|
+ Analyse descriptive univariée
|
|
|
+ variable : dictionnaire contenant la liste des variables à analyser de la forme :
|
|
|
+ Key = Nom de la variable, Value : type de variable : quantitative ou qualitative
|
|
|
+ axes : axes d'analyse de la variable, doivent être de type qualitative. Liste de variables.
|
|
|
+ """
|
|
|
+
|
|
|
+ # Sortie des résultats
|
|
|
+ resultats = {}
|
|
|
+
|
|
|
+ for variable, type_variable in variables.items():
|
|
|
+ if type_variable == 'qualitative':
|
|
|
+ resultats[variable] = self._analyse_univarie_qualitative(variable, axes)
|
|
|
+ elif type_variable == 'quantitative':
|
|
|
+ resultats[variable] = self._analyse_univarie_quantitative(variable, axes)
|
|
|
+
|
|
|
+ return(resultats)
|
19-0009Cse