Notebook de la séance 8

import pandas

df = pandas.read_excel("Donnees_M2_RD.xlsx")
df["reussi"] = (((df["Dist_A"] > df["Dist_B"]) & (df["Response"] == 2)) 
                 | ((df["Dist_A"] < df["Dist_B"]) & (df["Response"] == 1)))

Index à partir de séries

s123 = pandas.Series([1,2,3])
sabcd = pandas.Series(["A","B","C","D"])
pandas.MultiIndex.from_product(
    [s123, sabcd],
    names=["chiffres", "lettres"])

MultiIndex([(1, 'A'),
            (1, 'B'),
            (1, 'C'),
            (1, 'D'),
            (2, 'A'),
            (2, 'B'),
            (2, 'C'),
            (2, 'D'),
            (3, 'A'),
            (3, 'B'),
            (3, 'C'),
            (3, 'D')],
           names=['chiffres', 'lettres'])

Créer une liste à partir d’une liste existante

valeurs = [ 3, 4, 5, 6 , 7, 8 ]
valeurs_plus_3 = [ v+3 for v in valeurs ]
valeurs_plus_3

[6, 7, 8, 9, 10, 11]

df["Subject"].drop_duplicates()

0       P_ADI_331
400     P_ALM_345
800     P_AMY_346
1200    P_BAM_347
1600    P_BEH_340
2000    P_BLC_325
2399    P_BLR_321
2798    P_BOA_321
3197    P_BOC_342
3597    P_CAR_327
3995    P_CAV_333
4395    P_CON_336
4795    P_GAM_338
5195    P_GHM_334
5595    P_GRC_341
5995    P_GRF_322
6394    P_LAC_354
6794    P_LEG_335
7194    P_MOE_339
7594    P_ROS_336
7994    P_SOA_337
8394    P_TAI_343
8794    P_VAL_329
9194    P_VAR_330
Name: Subject, dtype: object

df["Name_A"].drop_duplicates()

0     0
1     1
2     4
3     2
17    3
Name: Name_A, dtype: int64

colonnes = [ "Subject", "Space", "Name_A", "Name_B", "Dist_A", "Dist_B", "Mode", "Side" ]
valeurs_colonnes = [ df[col].drop_duplicates() for col in colonnes ]

Fabrication d’un index par combinaison de ces valeurs

combinaisons = pandas.MultiIndex.from_product(
    valeurs_colonnes,
    names=colonnes)
combinaisons

MultiIndex([('P_ADI_331', 'E', 0, 2, 2, 4, 'Dic', 'D'),
            ('P_ADI_331', 'E', 0, 2, 2, 4, 'Dic', 'G'),
            ('P_ADI_331', 'E', 0, 2, 2, 4, 'Dio', 'D'),
            ('P_ADI_331', 'E', 0, 2, 2, 4, 'Dio', 'G'),
            ('P_ADI_331', 'E', 0, 2, 2, 1, 'Dic', 'D'),
            ('P_ADI_331', 'E', 0, 2, 2, 1, 'Dic', 'G'),
            ('P_ADI_331', 'E', 0, 2, 2, 1, 'Dio', 'D'),
            ('P_ADI_331', 'E', 0, 2, 2, 1, 'Dio', 'G'),
            ('P_ADI_331', 'E', 0, 2, 2, 2, 'Dic', 'D'),
            ('P_ADI_331', 'E', 0, 2, 2, 2, 'Dic', 'G'),
            ...
            ('P_VAR_330', 'I', 3, 0, 5, 2, 'Dio', 'D'),
            ('P_VAR_330', 'I', 3, 0, 5, 2, 'Dio', 'G'),
            ('P_VAR_330', 'I', 3, 0, 5, 3, 'Dic', 'D'),
            ('P_VAR_330', 'I', 3, 0, 5, 3, 'Dic', 'G'),
            ('P_VAR_330', 'I', 3, 0, 5, 3, 'Dio', 'D'),
            ('P_VAR_330', 'I', 3, 0, 5, 3, 'Dio', 'G'),
            ('P_VAR_330', 'I', 3, 0, 5, 5, 'Dic', 'D'),
            ('P_VAR_330', 'I', 3, 0, 5, 5, 'Dic', 'G'),
            ('P_VAR_330', 'I', 3, 0, 5, 5, 'Dio', 'D'),
            ('P_VAR_330', 'I', 3, 0, 5, 5, 'Dio', 'G')],
           names=['Subject', 'Space', 'Name_A', 'Name_B', 'Dist_A', 'Dist_B', 'Mode', 'Side'], length=120000)

df.loc[df["reussi"], "nombre_reussis"] = 1
df.loc[~ (df["reussi"]), "nombre_reussis"] = 0
df["essais"] = 1
df2 = df.set_index(colonnes)[["nombre_reussis", "essais"]]
df3 = df2.reindex(index=combinaisons, fill_value=0)
df3

Les combinaisons sans essai

df3[df3["essais"] == 0]

Ajout d’une colonne du nombre de combinaisons

df3["combinaisons"] = 1

Taux de combinaisons explorées par sujet

comb_subj = df3.groupby(by="Subject")["combinaisons"].sum()
essais_subj = df3.groupby(by="Subject")["essais"].sum()
essais_subj / comb_subj

Subject
P_ADI_331    0.0800
P_ALM_345    0.0800
P_AMY_346    0.0800
P_BAM_347    0.0800
P_BEH_340    0.0800
P_BLC_325    0.0798
P_BLR_321    0.0798
P_BOA_321    0.0798
P_BOC_342    0.0800
P_CAR_327    0.0796
P_CAV_333    0.0800
P_CON_336    0.0800
P_GAM_338    0.0800
P_GHM_334    0.0800
P_GRC_341    0.0800
P_GRF_322    0.0798
P_LAC_354    0.0800
P_LEG_335    0.0800
P_MOE_339    0.0800
P_ROS_336    0.0800
P_SOA_337    0.0800
P_TAI_343    0.0800
P_VAL_329    0.0800
P_VAR_330    0.0800
dtype: float64

Taux de combinaisons explorées par sujet et Mode

comb_subj = df3.groupby(by=["Subject","Mode"])["combinaisons"].sum()
essais_subj = df3.groupby(by=["Subject", "Mode"])["essais"].sum()
essais_subj / comb_subj

Subject    Mode
P_ADI_331  Dic     0.0800
           Dio     0.0800
P_ALM_345  Dic     0.0800
           Dio     0.0800
P_AMY_346  Dic     0.0800
           Dio     0.0800
P_BAM_347  Dic     0.0800
           Dio     0.0800
P_BEH_340  Dic     0.0800
           Dio     0.0800
P_BLC_325  Dic     0.0796
           Dio     0.0800
P_BLR_321  Dic     0.0796
           Dio     0.0800
P_BOA_321  Dic     0.0796
           Dio     0.0800
P_BOC_342  Dic     0.0800
           Dio     0.0800
P_CAR_327  Dic     0.0796
           Dio     0.0796
P_CAV_333  Dic     0.0800
           Dio     0.0800
P_CON_336  Dic     0.0800
           Dio     0.0800
P_GAM_338  Dic     0.0800
           Dio     0.0800
P_GHM_334  Dic     0.0800
           Dio     0.0800
P_GRC_341  Dic     0.0800
           Dio     0.0800
P_GRF_322  Dic     0.0796
           Dio     0.0800
P_LAC_354  Dic     0.0800
           Dio     0.0800
P_LEG_335  Dic     0.0800
           Dio     0.0800
P_MOE_339  Dic     0.0800
           Dio     0.0800
P_ROS_336  Dic     0.0800
           Dio     0.0800
P_SOA_337  Dic     0.0800
           Dio     0.0800
P_TAI_343  Dic     0.0800
           Dio     0.0800
P_VAL_329  Dic     0.0800
           Dio     0.0800
P_VAR_330  Dic     0.0800
           Dio     0.0800
dtype: float64