Schuhgröße – Geschlecht

import pandas as pd

datei = "synthetische_daten_3bis40_mit_BMI_Kategorien.csv"

df = pd.read_csv(datei, index_col="ID")

df.head()

	Alter_Jahre	Geschlecht	Körpergröße_cm	Gewicht_kg	BMI	BMI_Kategorie	Schuhgröße_EU
ID
1	13	w	160.2	51.1	19.9	Normalgewicht	38
2	7	m	128.1	23.6	14.4	Untergewichtig	31
3	4	w	100.7	16.5	16.3	Normalgewicht	25
4	11	w	148.9	39.5	17.8	Normalgewicht	36
5	9	w	129.7	30.1	17.9	Normalgewicht	31

from sklearn.model_selection import train_test_split

y = df["Geschlecht"]

type(y)

pandas.core.series.Series

df.columns

Index(['Alter_Jahre', 'Geschlecht', 'Körpergröße_cm', 'Gewicht_kg', 'BMI',
       'BMI_Kategorie', 'Schuhgröße_EU'],
      dtype='object')

X = df.drop(["Geschlecht" , "BMI_Kategorie" ] , axis=1)

X

	Alter_Jahre	Körpergröße_cm	Gewicht_kg	BMI	Schuhgröße_EU
ID
1	13	160.2	51.1	19.9	38
2	7	128.1	23.6	14.4	31
3	4	100.7	16.5	16.3	25
4	11	148.9	39.5	17.8	36
5	9	129.7	30.1	17.9	31
...	...	...	...	...	...
1496	25	170.6	57.0	19.6	41
1497	30	164.3	54.3	20.1	39
1498	35	167.0	69.4	24.9	40
1499	36	168.8	62.1	21.8	40
1500	33	179.3	81.7	25.4	43

1500 rows × 5 columns

X.describe()

	Alter_Jahre	Körpergröße_cm	Gewicht_kg	BMI	Schuhgröße_EU
count	1500.000000	1500.000000	1500.000000	1500.000000	1500.000000
mean	18.227333	153.753067	52.699200	20.692067	36.854667
std	10.560587	25.734906	23.107748	4.288830	5.804711
min	3.000000	76.200000	7.800000	8.500000	19.000000
25%	9.000000	135.700000	30.900000	17.100000	33.000000
50%	17.000000	162.950000	57.700000	20.900000	39.000000
75%	27.000000	173.000000	71.200000	24.000000	41.000000
max	40.000000	197.300000	111.500000	33.100000	47.000000

from sklearn.model_selection import train_test_split

train_X, val_X, train_y, val_y = train_test_split(X, y, random_state = 0)

train_X.shape

(1125, 5)

val_X.shape

(375, 5)

from sklearn.linear_model import LogisticRegression

reg = LogisticRegression() # regression in weiterem Sinn: auch Klassen

reg.fit(train_X, train_y)

LogisticRegression()

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

val_X

	Alter_Jahre	Körpergröße_cm	Gewicht_kg	BMI	Schuhgröße_EU
ID
472	16	162.0	55.6	21.2	39
10	13	156.3	46.9	19.2	37
1500	33	179.3	81.7	25.4	43
55	10	147.6	31.2	14.3	35
1412	27	184.0	75.2	22.2	44
...	...	...	...	...	...
76	5	105.4	16.3	14.7	26
482	13	159.5	56.2	22.1	38
1351	38	169.7	68.8	23.9	40
749	10	136.8	34.2	18.3	33
1363	35	167.0	77.3	27.7	40

375 rows × 5 columns

predict_y = reg.predict( val_X ) 

len(predict_y), len(val_y)

(375, 375)

predict_y[:10]

array(['w', 'w', 'm', 'm', 'm', 'm', 'm', 'w', 'm', 'm'], dtype=object)

list(val_y[:10])

['w', 'm', 'm', 'm', 'm', 'm', 'm', 'w', 'w', 'm']

"""
differenz = []
for i in range(0,10):
    v = list(val_y)[i]
    p = list(predict_y)[i]
    
    # print( f"val: {v}, predict: {p}")
    differenz.append( abs(v-p) )
differenz
MAE = sum( differenz ) / len( differenz )
MAE
"""

'\ndifferenz = []\nfor i in range(0,10):\n    v = list(val_y)[i]\n    p = list(predict_y)[i]\n\n    # print( f"val: {v}, predict: {p}")\n    differenz.append( abs(v-p) )\ndifferenz\nMAE = sum( differenz ) / len( differenz )\nMAE\n'

cm = { "m" : { "m": 0, "w": 0 },
       "w":  { "m": 0, "w": 0 }}

for i in range(len( val_y )):

    v = list(val_y)[i]
    p = list(predict_y)[i]
    
    #print(v ,  p )
    cm[ v ][ p ] += 1
cm

{'m': {'m': 105, 'w': 72}, 'w': {'m': 49, 'w': 149}}

Diskussion siehe https://en.wikipedia.org/wiki/Confusion_matrix