def max (n1, n2):
if n1 > n2:
print(n1)
elif n2 > n1:
print(n2)
else:
print("Son iguales")
max(799,1)
799
def max_de_tres (n1, n2, n3):
if n1 == n2 and n2 == n3 and n1 == n3:
print('Los tres numeros son iguales')
elif n1 >= n2 and n1 >= n3:
print(n1)
elif n2 >= n1 and n2 >= n3:
print(n2)
elif n3 >= n1 and n3 >= n2:
print(n3)
else:
print("Ingrese otras vez los numeros, por favor")
max_de_tres(1,0,0)
1
def largor (entrada):
contador = 0
for item in entrada:
contador += 1
return contador
largor("Onomatopeya")
def vocal (caracter):
l_vocales = ['a','e','i','o','u']
if caracter in l_vocales:
print(f"El caracter '{caracter}' es una vocal")
else:
print("No es una vocal")
vocal('u')
El caracter 'u' es una vocal
def suma(lista):
suma=0
for item in range(len(lista)):
suma=suma+lista[item]
return suma
suma([1,1,1,10,10,10])
def reverso (cadena):
reversada = cadena[::-1]
print(reversada)
cadena = 'estoy probando'
reverso(cadena)
odnaborp yotse
def generar_n_caracteres(n, carcater):
resultado = carcater*n
print(resultado)
generar_n_caracteres(5,'z')
zzzzz
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error
#from sklearn import datasets
# Importar la base
#boston_data = datasets.load_boston()
data_url = "http://lib.stat.cmu.edu/datasets/boston"
raw_df = pd.read_csv(data_url, sep="\s+", skiprows=22, header=None)
data = np.hstack([raw_df.values[::2, :], raw_df.values[1::2, :2]])
target = raw_df.values[1::2, 2]
#df = pd.DataFrame(boston_data.data, columns=boston_data.feature_names)
feature_names = ['CRIM', 'ZN', 'INDUS', 'CHAS', 'NOX', 'RM', 'AGE', 'DIS', 'RAD', 'TAX', 'PTRATIO', 'B', 'LSTAT']
df = pd.DataFrame(data, columns=feature_names)
df.head()
CRIMfloat64
ZNfloat64
0
0.00632
18.0
1
0.02731
0.0
2
0.02729
0.0
3
0.03237
0.0
4
0.06905
0.0
#df['MEDV'] = boston_data.target
df['MEDV'] = target
df.head()
CRIMfloat64
ZNfloat64
0
0.00632
18.0
1
0.02731
0.0
2
0.02729
0.0
3
0.03237
0.0
4
0.06905
0.0
df.isnull()
CRIMbool
False100%
ZNbool
False100%
0
False
False
1
False
False
2
False
False
3
False
False
4
False
False
5
False
False
6
False
False
7
False
False
8
False
False
9
False
False
df.isnull().sum()
# columnas a usar en el dataframe
categorica_test = ["blanca", "roja", "verde"]
continua_test = [1, 2, 3]
# pásalas a formato dataframe
df_test = pd.DataFrame(list(zip(categorica_test, continua_test)),
columns =['color', 'valor'])
df_test
colorobject
valorint64
0
blanca
1
1
roja
2
2
verde
3
pd.get_dummies(df_test,columns=['color'])
valorint64
color_blancauint8
0
1
1
1
2
0
2
3
0
precios = df.MEDV
# Algunas estadísticas básicas
precio_minimo = np.min(precios)
# Alternativa usando Pandas
# precio_minimo = precios.min()
precio_maximo = np.max(precios)
# Alternativa usando Pandas
# precio_maximo = precios.max()
precio_promedio = np.mean(precios)
# Alternativa usando Pandas
# precio_promedio = precios.mean()
precio_mediano = np.median(precios)
# Alternativa usando Pandas
# precio_mediano = precios.median()
precio_desvest = np.std(precios)
# Alternativa usando Pandas
# precio_desvest = precios.std(ddof=0)
# Otras estadísticas
cuartil_1 = np.percentile(precios, 25)
cuartil_3 = np.percentile(precios, 75)
inter_cuartil = cuartil_3 - cuartil_1
# Muestra las estadísticas
print ("Estadísticas para el conjunto de datos de Boston-Dataset:\n")
print ("Precio Mínimo: ${:,.2f}".format(precio_minimo))
print ("Precio Máximo: ${:,.2f}".format(precio_maximo))
print ("Precio Promedio: ${:,.2f}".format(precio_promedio))
print ("Precio Mediano: ${:,.2f}".format(precio_mediano))
print ("Desviación Estándar de Precios: ${:,.2f}".format(precio_desvest))
print ("Primer Cuartil de Precios: ${:,.2f}".format(cuartil_1))
print ("Tercer Cuartil de Precios: ${:,.2f}".format(cuartil_3))
print ("Rango Intercuantílico de Precios: ${:,.2f}".format(inter_cuartil))
Estadísticas para el conjunto de datos de Boston-Dataset:
Precio Mínimo: $5.00
Precio Máximo: $50.00
Precio Promedio: $22.53
Precio Mediano: $21.20
Desviación Estándar de Precios: $9.19
Primer Cuartil de Precios: $17.02
Tercer Cuartil de Precios: $25.00
Rango Intercuantílico de Precios: $7.98
sns.set(rc={'figure.figsize':(11.7,8.27)})
sns.histplot(df['MEDV'], bins=30)
plt.show()
plt.plot(df.B, df.MEDV, 'o') # 'o' muestra scatter (por defecto es línea)
x = "INDUS"
y = "MEDV"
sns.lmplot(x=x,y=y,data=df)
precios = df.MEDV
columnas = ["LSTAT", "RM", "INDUS"]
for i, col in enumerate(columnas):
sns.lmplot(x=col,y="MEDV",data=df,fit_reg=True)
correlation_matrix = df.corr().round(2)
# annot = True; para imprimir los valores en cada cuadro
sns.heatmap(data=correlation_matrix, annot=True)
# renombra las variables para usar las convenciones de X, y
X_rooms = df.RM # única variables explicativa
y_price = df.MEDV # variable a predecir
X_rooms
X_rooms.shape
# es mejor usar arreglos tamaño (n, 1)
X_rooms = np.array(X_rooms).reshape(-1,1)
y_price = np.array(y_price).reshape(-1,1)
print(X_rooms.shape)
print(y_price.shape)
(506, 1)
(506, 1)
# en una sóla línea de código se obtienen los cuatros segmentos
X_train_1, X_test_1, Y_train_1, Y_test_1 = train_test_split(X_rooms, y_price, test_size = 0.2, random_state=5)
# asegúrate que las dimensiones de los datos son las esperadas
print(X_train_1.shape)
print(X_test_1.shape)
print(Y_train_1.shape)
print(Y_test_1.shape)
(404, 1)
(102, 1)
(404, 1)
(102, 1)
# instancia el modelo
reg_1 = LinearRegression()
reg_1.fit(X_train_1, Y_train_1) # estima el modelo siempre con datos de entrenamiento
y_train_predict_1 = reg_1.predict(X_train_1) # genera las predicciones
# RMSE en numpy usa sólo los valores reales y las predicciones
rmse = (np.sqrt(mean_squared_error(Y_train_1, y_train_predict_1)))
# en scikitlearn se estima en una sóla línea las predicciones
r2 = round(reg_1.score(X_train_1, Y_train_1),2)
print("Desempeño del modelo para los datos de entrenamiento")
print("--------------------------------------")
print('RMSE: {}'.format(rmse))
print('R2: {}'.format(r2))
Desempeño del modelo para los datos de entrenamiento
--------------------------------------
RMSE: 6.972277149440585
R2: 0.43
# evaluación del modelo para datos de test
y_pred_1 = reg_1.predict(X_test_1)
rmse = (np.sqrt(mean_squared_error(Y_test_1, y_pred_1)))
r2 = round(reg_1.score(X_test_1, Y_test_1),2)
print("Desempeño del modelo para los datos de test")
print("--------------------------------------")
print("RMSE: {}".format(rmse))
print("R2: {}".format(r2))
Desempeño del modelo para los datos de test
--------------------------------------
RMSE: 4.895963186952216
R2: 0.69
# 1. Definiendo variables
# "X" -> conjunto de variables explicativas
# "y" -> variable a predecir (precios)
x = df.drop(columns=['MEDV'])
y = df.MEDV
# 2. Segmentación de Train y Test
x_train, x_test, y_train, y_test = train_test_split(x, y_price, test_size = 0.2, random_state=42)
# 3. Instanciar la regresión lineal y ajustar el modelo
reg_all = LinearRegression()
reg_all.fit(x_train, y_train)
# 4. Generar las predicciones sobre el conjunto X de entrenamiento
y_train_predict_all = reg_all.predict(x_train)
# 4a. Calcular el RMSE y R2
rmse = (np.sqrt(mean_squared_error(y_train, y_train_predict_all)))
r2 = round(reg_all.score(x_train, y_train),2)
print("Desempeño del modelo para los datos de entrenamiento")
print("--------------------------------------")
print('RMSE: {}'.format(rmse))
print('R2: {}'.format(r2))
Desempeño del modelo para los datos de entrenamiento
--------------------------------------
RMSE: 4.6520331848801675
R2: 0.75
# 5. Generar las predicciones sobre el conjunto X de test
y_pred = reg_all.predict(x_test)
rmse = (np.sqrt(mean_squared_error(y_test, y_pred)))
r2 = round(reg_all.score(x_test, y_test),2)
print("Desempeño del modelo para los datos de test")
print("--------------------------------------")
print("RMSE: {}".format(rmse))
print("R2: {}".format(r2))
Desempeño del modelo para los datos de test
--------------------------------------
RMSE: 4.928602182665323
R2: 0.67
# pasa los precios reales de test y los pronósticos para graficarlos
plt.scatter(y_pred, y_test) # dos argumentos
# estas 5 líneas de código no necesitas modificarlas
plt.xlabel("Precio Real ($1000)")
plt.ylabel("Precio Pronosticado ($1000)")
plt.title("Precios Reales vs. Precios Pronosticados de Casas")
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sn
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelEncoder
from sklearn.preprocessing import MinMaxScaler
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import confusion_matrix
tc_data = pd.read_csv('crx.data', header=None)
tc_data.head()
0object
1object
0
b
30.83
1
a
58.67
2
a
24.50
3
b
27.83
4
b
20.17
# diccionario con nombres
nuevas_variables = {
0: "genero",
1: "edad",
2: "deuda",
3: "estado_civil",
4: "cliente",
5: "educacion",
6: "etnia",
7: "años_empleado",
8: "no_pago",
9: "empleado",
10: "score_crediticio",
11: "licencia_conduccion",
12: "ciudadania",
13: "zip",
14: "ingreso",
15: "tarjeta_aprobada"
}
# usa el diccionario para renombrar las variables
tc_data = tc_data.rename(columns = nuevas_variables)
tc_data.head()
generoobject
edadobject
0
b
30.83
1
a
58.67
2
a
24.50
3
b
27.83
4
b
20.17
tc_data.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 690 entries, 0 to 689
Data columns (total 16 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 genero 690 non-null object
1 edad 690 non-null object
2 deuda 690 non-null float64
3 estado_civil 690 non-null object
4 cliente 690 non-null object
5 educacion 690 non-null object
6 etnia 690 non-null object
7 años_empleado 690 non-null float64
8 no_pago 690 non-null object
9 empleado 690 non-null object
10 score_crediticio 690 non-null int64
11 licencia_conduccion 690 non-null object
12 ciudadania 690 non-null object
13 zip 690 non-null object
14 ingreso 690 non-null int64
15 tarjeta_aprobada 690 non-null object
dtypes: float64(2), int64(2), object(12)
memory usage: 86.4+ KB
# imprime todos los valores únicos de cada columna
for col in tc_data.columns:
print(col)
print(pd.unique(tc_data[col]))
print()
genero
['b' 'a' '?']
edad
['30.83' '58.67' '24.50' '27.83' '20.17' '32.08' '33.17' '22.92' '54.42'
'42.50' '22.08' '29.92' '38.25' '48.08' '45.83' '36.67' '28.25' '23.25'
'21.83' '19.17' '25.00' '47.75' '27.42' '41.17' '15.83' '47.00' '56.58'
'57.42' '42.08' '29.25' '42.00' '49.50' '36.75' '22.58' '27.25' '23.00'
'27.75' '54.58' '34.17' '28.92' '29.67' '39.58' '56.42' '54.33' '41.00'
'31.92' '41.50' '23.92' '25.75' '26.00' '37.42' '34.92' '34.25' '23.33'
'23.17' '44.33' '35.17' '43.25' '56.75' '31.67' '23.42' '20.42' '26.67'
'36.00' '25.50' '19.42' '32.33' '34.83' '38.58' '44.25' '44.83' '20.67'
'34.08' '21.67' '21.50' '49.58' '27.67' '39.83' '?' '37.17' '25.67'
'34.00' '49.00' '62.50' '31.42' '52.33' '28.75' '28.58' '22.50' '28.50'
'37.50' '35.25' '18.67' '54.83' '40.92' '19.75' '29.17' '24.58' '33.75'
'25.42' '37.75' '52.50' '57.83' '20.75' '39.92' '24.75' '44.17' '23.50'
'47.67' '22.75' '34.42' '28.42' '67.75' '47.42' '36.25' '32.67' '48.58'
'33.58' '18.83' '26.92' '31.25' '56.50' '43.00' '22.33' '32.83' '40.33'
'30.50' '52.83' '46.67' '58.33' '37.33' '23.08' '32.75' '68.67' '28.00'
'44.00' '25.08' '32.00' '60.58' '40.83' '19.33' '41.33' '56.00' '49.83'
'22.67' '27.00' '26.08' '18.42' '21.25' '57.08' '22.42' '48.75' '40.00'
'40.58' '28.67' '33.08' '21.33' '41.75' '34.50' '48.17' '27.58' '24.08'
'24.83' '36.33' '35.42' '71.58' '39.50' '39.33' '24.33' '60.08' '55.92'
'53.92' '18.92' '50.08' '65.42' '17.58' '18.08' '19.67' '25.17' '33.50'
'58.42' '26.17' '42.83' '38.17' '20.50' '48.25' '28.33' '18.75' '18.50'
'45.00' '40.25' '41.42' '17.83' '18.17' '20.00' '52.17' '50.75' '17.08'
'18.33' '59.67' '18.00' '37.58' '30.67' '18.58' '16.25' '21.17' '17.67'
'16.50' '29.50' '21.75' '18.25' '35.75' '16.08' '69.17' '32.92' '16.33'
'22.17' '57.58' '15.92' '31.75' '19.00' '17.50' '33.67' '30.17' '33.25'
'25.25' '34.75' '47.33' '39.08' '42.75' '38.92' '62.75' '32.25' '26.75'
'63.33' '30.75' '16.00' '19.50' '32.42' '30.25' '26.83' '16.92' '24.42'
'39.42' '23.58' '21.42' '33.00' '26.33' '26.25' '28.17' '20.83' '43.17'
'56.83' '15.17' '29.83' '31.00' '51.92' '69.50' '19.58' '22.25' '38.42'
'26.58' '35.00' '29.42' '49.17' '51.83' '58.58' '53.33' '27.17' '25.92'
'30.58' '17.25' '27.33' '36.50' '29.75' '52.42' '36.17' '34.58' '21.92'
'36.58' '31.08' '30.42' '21.08' '17.42' '39.17' '26.50' '17.33' '23.75'
'34.67' '74.83' '45.33' '47.25' '24.17' '39.25' '39.00' '64.08' '31.33'
'21.00' '13.75' '46.00' '20.25' '60.92' '30.00' '22.83' '45.17' '41.58'
'55.75' '25.33' '31.83' '33.92' '24.92' '80.25' '30.08' '48.33' '76.75'
'51.33' '41.92' '29.58' '32.17' '51.42' '42.17' '43.08' '59.50' '65.17'
'20.33' '48.50' '28.08' '73.42' '51.58' '38.67' '46.08' '20.08' '42.25'
'16.17' '47.83' '22.00' '38.33' '25.58' '21.58' '36.08' '38.75' '35.58'
'31.58' '15.75' '17.92' '30.33' '47.17' '25.83' '50.25' '36.42']
deuda
[ 0. 4.46 0.5 1.54 5.625 4. 1.04 11.585 4.915 0.83
1.835 6. 6.04 10.5 4.415 0.875 5.875 0.25 8.585 11.25
1. 8. 14.5 6.5 0.585 13. 18.5 8.5 14.79 9.79
7.585 5.125 10.75 1.5 1.585 11.75 9.415 9.17 15. 1.415
13.915 28. 6.75 2.04 0.665 2.5 3. 11.625 4.5 12.25
16.165 0.79 0.835 4.25 0.375 25.125 7.5 5. 7. 5.29
1.165 9.75 19. 3.5 0.625 2.21 12.75 15.5 1.375 3.54
11. 1.75 16.5 12. 2.25 0.75 12.5 1.25 1.125 7.04
10.335 6.21 6.665 9. 5.5 0.54 2.75 9.5 13.5 3.75
16. 0.29 1.665 7.54 0.46 10. 11.5 3.04 2. 0.08
1.71 3.25 2.54 13.585 8.665 9.25 8.17 2.335 19.5 5.665
4.625 0.205 0.96 4.04 5.04 3.165 7.625 10.04 10.25 2.125
9.335 6.625 2.71 9.625 12.54 9.54 8.46 13.75 21. 10.125
25.085 0.21 21.5 11.125 11.045 1.335 0.085 1.21 0.165 5.71
5.415 12.625 0.58 0.415 2.415 0.335 3.125 12.125 2.875 13.665
26.335 10.29 1.29 22. 0.125 1.085 4.085 4.71 6.165 4.585
11.46 14.585 0.17 1.625 2.085 5.085 8.125 2.835 1.79 0.705
2.165 2.29 18.125 3.085 11.665 4.125 1.08 13.335 11.835 4.79
9.96 7.08 25.21 0.67 3.79 22.29 3.335 0.42 1.46 0.04
12.33 12.335 0.915 14. 17.75 20. 5.25 4.165 10.915 4.75
10.415 7.835 0.71 2.46 9.585 3.625 2.665 5.835 12.835 10.665
7.25 10.21 3.29 10.085 3.375]
estado_civil
['u' 'y' '?' 'l']
cliente
['g' 'p' '?' 'gg']
educacion
['w' 'q' 'm' 'r' 'cc' 'k' 'c' 'd' 'x' 'i' 'e' 'aa' 'ff' 'j' '?']
etnia
['v' 'h' 'bb' 'ff' 'j' 'z' '?' 'o' 'dd' 'n']
años_empleado
[ 1.25 3.04 1.5 3.75 1.71 2.5 6.5 0.04 3.96 3.165
2.165 4.335 1. 5. 0.25 0.96 3.17 0.665 0.75 0.835
7.875 3.085 0.5 5.165 15. 7. 5.04 7.96 7.585 0.415
2. 1.835 14.415 4.5 5.335 8.625 28.5 2.625 0.125 6.04
3.5 0.165 0.875 1.75 0. 7.415 0.085 5.75 6. 3.
1.585 4.29 1.54 1.46 1.625 12.5 13.5 10.75 0.375 0.585
0.455 4. 8.5 9.46 2.25 10. 0.795 1.375 1.29 11.5
6.29 14. 0.335 1.21 7.375 7.5 3.25 13. 5.5 4.25
0.625 5.085 2.75 2.375 8. 1.085 2.54 4.165 1.665 11.
9. 1.335 1.415 1.96 2.585 5.125 15.5 0.71 5.665 18.
5.25 8.665 2.29 20. 2.46 13.875 2.085 4.58 2.71 2.04
0.29 4.75 0.46 0.21 0.54 3.335 2.335 1.165 2.415 2.79
4.625 1.04 6.75 1.875 16. 12.75 5.375 2.125 17.5 3.125
0.79 8.29 ]
no_pago
['t' 'f']
empleado
['t' 'f']
score_crediticio
[ 1 6 0 5 7 10 3 17 2 9 8 15 11 12 40 23 4 20 67 14 16 13 19]
licencia_conduccion
['f' 't']
ciudadania
['g' 's' 'p']
zip
['00202' '00043' '00280' '00100' '00120' '00360' '00164' '00080' '00180'
'00052' '00128' '00260' '00000' '00320' '00396' '00096' '00200' '00300'
'00145' '00500' '00168' '00434' '00583' '00030' '00240' '00070' '00455'
'00311' '00216' '00491' '00400' '00239' '00160' '00711' '00250' '00520'
'00515' '00420' '?' '00980' '00443' '00140' '00094' '00368' '00288'
'00928' '00188' '00112' '00171' '00268' '00167' '00075' '00152' '00176'
'00329' '00212' '00410' '00274' '00375' '00408' '00350' '00204' '00040'
'00181' '00399' '00440' '00093' '00060' '00395' '00393' '00021' '00029'
'00102' '00431' '00370' '00024' '00020' '00129' '00510' '00195' '00144'
'00380' '00049' '00050' '00381' '00150' '00117' '00056' '00211' '00230'
'00156' '00022' '00228' '00519' '00253' '00487' '00220' '00088' '00073'
'00121' '00470' '00136' '00132' '00292' '00154' '00272' '00340' '00108'
'00720' '00450' '00232' '00170' '01160' '00411' '00460' '00348' '00480'
'00640' '00372' '00276' '00221' '00352' '00141' '00178' '00600' '00550'
'02000' '00225' '00210' '00110' '00356' '00045' '00062' '00092' '00174'
'00017' '00086' '00454' '00254' '00028' '00263' '00333' '00312' '00290'
'00371' '00099' '00252' '00760' '00560' '00130' '00523' '00680' '00163'
'00208' '00383' '00330' '00422' '00840' '00432' '00032' '00186' '00303'
'00349' '00224' '00369' '00076' '00231' '00309' '00416' '00465' '00256']
ingreso
[ 0 560 824 3 31285 1349 314 1442 200 2690
245 1208 1260 11 10000 5000 4000 35 713 551
500 300 221 2283 100 15 284 1236 5800 730
400 50000 456 15108 2954 2 20 27 225 1
38 5 130 147 210 11202 1332 50 258 567
1000 2510 809 610 150 51100 367 600 247 375
278 827 2072 582 2300 3065 2200 6 1602 2184
3376 2000 7544 10561 837 11177 639 2028 1065 540
158 15000 3000 3257 1655 1430 7 790 396 678
1187 6590 168 1270 1210 742 8851 7059 1704 857
6700 2503 9800 196 14 26726 18027 99 444 1200
2010 13 120 32 722 40 484 204 98 5552
105 2803 126 4 21 173 10 25 42 100000
113 8 44 2732 179 16 1062 251 228 67
12 122 4208 1300 112 1110 1004 286 4500 1212
195 87 17 184 140 18 146 22 55 70
60 1058 769 5200 19 316 350 3552 687 1950
53 41 33 80 351 2100 475 892 4607 2206
5860 28 1391 2279 591 960 690 234 800 990
2197 90 340 347 327 4071 109 1249 134 1344
321 948 2079 2384 458 5298 162 1583 58 59
1400 1465 8000 4700 1097 3290 13212 5777 5124 23
4159 918 768 283 108 9 68 587 141 501
160 390 154 117 246 237 364 537 394 750]
tarjeta_aprobada
['+' '-']
# Reemplaza los '?'s con NaN
tc_data = tc_data.replace('?', np.nan)
# Imputar los no reportados con el promedio
tc_data.fillna(tc_data.mean(), inplace=True)
# Verifica que las variables numéricas no tengan NAN
print(tc_data.isnull().sum())
genero 12
edad 12
deuda 0
estado_civil 6
cliente 6
educacion 9
etnia 9
años_empleado 0
no_pago 0
empleado 0
score_crediticio 0
licencia_conduccion 0
ciudadania 0
zip 13
ingreso 0
tarjeta_aprobada 0
dtype: int64
# imputar con el valor más frecuente los valores no reportados en las variables categóricas
for col in tc_data.columns:
if tc_data[col].dtypes == 'object':
tc_data[col].fillna(tc_data[col].mode()[0], inplace=True)
# Verifica que las variables numéricas no tengan NAN
print(tc_data.isnull().sum())
genero 0
edad 0
deuda 0
estado_civil 0
cliente 0
educacion 0
etnia 0
años_empleado 0
no_pago 0
empleado 0
score_crediticio 0
licencia_conduccion 0
ciudadania 0
zip 0
ingreso 0
tarjeta_aprobada 0
dtype: int64
# 2da revision: imprime todos los valores únicos de cada columna
for col in tc_data.columns:
print(col)
print(pd.unique(tc_data[col]))
print()
genero
['b' 'a']
edad
['30.83' '58.67' '24.50' '27.83' '20.17' '32.08' '33.17' '22.92' '54.42'
'42.50' '22.08' '29.92' '38.25' '48.08' '45.83' '36.67' '28.25' '23.25'
'21.83' '19.17' '25.00' '47.75' '27.42' '41.17' '15.83' '47.00' '56.58'
'57.42' '42.08' '29.25' '42.00' '49.50' '36.75' '22.58' '27.25' '23.00'
'27.75' '54.58' '34.17' '28.92' '29.67' '39.58' '56.42' '54.33' '41.00'
'31.92' '41.50' '23.92' '25.75' '26.00' '37.42' '34.92' '34.25' '23.33'
'23.17' '44.33' '35.17' '43.25' '56.75' '31.67' '23.42' '20.42' '26.67'
'36.00' '25.50' '19.42' '32.33' '34.83' '38.58' '44.25' '44.83' '20.67'
'34.08' '21.67' '21.50' '49.58' '27.67' '39.83' '22.67' '37.17' '25.67'
'34.00' '49.00' '62.50' '31.42' '52.33' '28.75' '28.58' '22.50' '28.50'
'37.50' '35.25' '18.67' '54.83' '40.92' '19.75' '29.17' '24.58' '33.75'
'25.42' '37.75' '52.50' '57.83' '20.75' '39.92' '24.75' '44.17' '23.50'
'47.67' '22.75' '34.42' '28.42' '67.75' '47.42' '36.25' '32.67' '48.58'
'33.58' '18.83' '26.92' '31.25' '56.50' '43.00' '22.33' '32.83' '40.33'
'30.50' '52.83' '46.67' '58.33' '37.33' '23.08' '32.75' '68.67' '28.00'
'44.00' '25.08' '32.00' '60.58' '40.83' '19.33' '41.33' '56.00' '49.83'
'27.00' '26.08' '18.42' '21.25' '57.08' '22.42' '48.75' '40.00' '40.58'
'28.67' '33.08' '21.33' '41.75' '34.50' '48.17' '27.58' '24.08' '24.83'
'36.33' '35.42' '71.58' '39.50' '39.33' '24.33' '60.08' '55.92' '53.92'
'18.92' '50.08' '65.42' '17.58' '18.08' '19.67' '25.17' '33.50' '58.42'
'26.17' '42.83' '38.17' '20.50' '48.25' '28.33' '18.75' '18.50' '45.00'
'40.25' '41.42' '17.83' '18.17' '20.00' '52.17' '50.75' '17.08' '18.33'
'59.67' '18.00' '37.58' '30.67' '18.58' '16.25' '21.17' '17.67' '16.50'
'29.50' '21.75' '18.25' '35.75' '16.08' '69.17' '32.92' '16.33' '22.17'
'57.58' '15.92' '31.75' '19.00' '17.50' '33.67' '30.17' '33.25' '25.25'
'34.75' '47.33' '39.08' '42.75' '38.92' '62.75' '32.25' '26.75' '63.33'
'30.75' '16.00' '19.50' '32.42' '30.25' '26.83' '16.92' '24.42' '39.42'
'23.58' '21.42' '33.00' '26.33' '26.25' '28.17' '20.83' '43.17' '56.83'
'15.17' '29.83' '31.00' '51.92' '69.50' '19.58' '22.25' '38.42' '26.58'
'35.00' '29.42' '49.17' '51.83' '58.58' '53.33' '27.17' '25.92' '30.58'
'17.25' '27.33' '36.50' '29.75' '52.42' '36.17' '34.58' '21.92' '36.58'
'31.08' '30.42' '21.08' '17.42' '39.17' '26.50' '17.33' '23.75' '34.67'
'74.83' '45.33' '47.25' '24.17' '39.25' '39.00' '64.08' '31.33' '21.00'
'13.75' '46.00' '20.25' '60.92' '30.00' '22.83' '45.17' '41.58' '55.75'
'25.33' '31.83' '33.92' '24.92' '80.25' '30.08' '48.33' '76.75' '51.33'
'41.92' '29.58' '32.17' '51.42' '42.17' '43.08' '59.50' '65.17' '20.33'
'48.50' '28.08' '73.42' '51.58' '38.67' '46.08' '20.08' '42.25' '16.17'
'47.83' '22.00' '38.33' '25.58' '21.58' '36.08' '38.75' '35.58' '31.58'
'15.75' '17.92' '30.33' '47.17' '25.83' '50.25' '36.42']
deuda
[ 0. 4.46 0.5 1.54 5.625 4. 1.04 11.585 4.915 0.83
1.835 6. 6.04 10.5 4.415 0.875 5.875 0.25 8.585 11.25
1. 8. 14.5 6.5 0.585 13. 18.5 8.5 14.79 9.79
7.585 5.125 10.75 1.5 1.585 11.75 9.415 9.17 15. 1.415
13.915 28. 6.75 2.04 0.665 2.5 3. 11.625 4.5 12.25
16.165 0.79 0.835 4.25 0.375 25.125 7.5 5. 7. 5.29
1.165 9.75 19. 3.5 0.625 2.21 12.75 15.5 1.375 3.54
11. 1.75 16.5 12. 2.25 0.75 12.5 1.25 1.125 7.04
10.335 6.21 6.665 9. 5.5 0.54 2.75 9.5 13.5 3.75
16. 0.29 1.665 7.54 0.46 10. 11.5 3.04 2. 0.08
1.71 3.25 2.54 13.585 8.665 9.25 8.17 2.335 19.5 5.665
4.625 0.205 0.96 4.04 5.04 3.165 7.625 10.04 10.25 2.125
9.335 6.625 2.71 9.625 12.54 9.54 8.46 13.75 21. 10.125
25.085 0.21 21.5 11.125 11.045 1.335 0.085 1.21 0.165 5.71
5.415 12.625 0.58 0.415 2.415 0.335 3.125 12.125 2.875 13.665
26.335 10.29 1.29 22. 0.125 1.085 4.085 4.71 6.165 4.585
11.46 14.585 0.17 1.625 2.085 5.085 8.125 2.835 1.79 0.705
2.165 2.29 18.125 3.085 11.665 4.125 1.08 13.335 11.835 4.79
9.96 7.08 25.21 0.67 3.79 22.29 3.335 0.42 1.46 0.04
12.33 12.335 0.915 14. 17.75 20. 5.25 4.165 10.915 4.75
10.415 7.835 0.71 2.46 9.585 3.625 2.665 5.835 12.835 10.665
7.25 10.21 3.29 10.085 3.375]
estado_civil
['u' 'y' 'l']
cliente
['g' 'p' 'gg']
educacion
['w' 'q' 'm' 'r' 'cc' 'k' 'c' 'd' 'x' 'i' 'e' 'aa' 'ff' 'j']
etnia
['v' 'h' 'bb' 'ff' 'j' 'z' 'o' 'dd' 'n']
años_empleado
[ 1.25 3.04 1.5 3.75 1.71 2.5 6.5 0.04 3.96 3.165
2.165 4.335 1. 5. 0.25 0.96 3.17 0.665 0.75 0.835
7.875 3.085 0.5 5.165 15. 7. 5.04 7.96 7.585 0.415
2. 1.835 14.415 4.5 5.335 8.625 28.5 2.625 0.125 6.04
3.5 0.165 0.875 1.75 0. 7.415 0.085 5.75 6. 3.
1.585 4.29 1.54 1.46 1.625 12.5 13.5 10.75 0.375 0.585
0.455 4. 8.5 9.46 2.25 10. 0.795 1.375 1.29 11.5
6.29 14. 0.335 1.21 7.375 7.5 3.25 13. 5.5 4.25
0.625 5.085 2.75 2.375 8. 1.085 2.54 4.165 1.665 11.
9. 1.335 1.415 1.96 2.585 5.125 15.5 0.71 5.665 18.
5.25 8.665 2.29 20. 2.46 13.875 2.085 4.58 2.71 2.04
0.29 4.75 0.46 0.21 0.54 3.335 2.335 1.165 2.415 2.79
4.625 1.04 6.75 1.875 16. 12.75 5.375 2.125 17.5 3.125
0.79 8.29 ]
no_pago
['t' 'f']
empleado
['t' 'f']
score_crediticio
[ 1 6 0 5 7 10 3 17 2 9 8 15 11 12 40 23 4 20 67 14 16 13 19]
licencia_conduccion
['f' 't']
ciudadania
['g' 's' 'p']
zip
['00202' '00043' '00280' '00100' '00120' '00360' '00164' '00080' '00180'
'00052' '00128' '00260' '00000' '00320' '00396' '00096' '00200' '00300'
'00145' '00500' '00168' '00434' '00583' '00030' '00240' '00070' '00455'
'00311' '00216' '00491' '00400' '00239' '00160' '00711' '00250' '00520'
'00515' '00420' '00980' '00443' '00140' '00094' '00368' '00288' '00928'
'00188' '00112' '00171' '00268' '00167' '00075' '00152' '00176' '00329'
'00212' '00410' '00274' '00375' '00408' '00350' '00204' '00040' '00181'
'00399' '00440' '00093' '00060' '00395' '00393' '00021' '00029' '00102'
'00431' '00370' '00024' '00020' '00129' '00510' '00195' '00144' '00380'
'00049' '00050' '00381' '00150' '00117' '00056' '00211' '00230' '00156'
'00022' '00228' '00519' '00253' '00487' '00220' '00088' '00073' '00121'
'00470' '00136' '00132' '00292' '00154' '00272' '00340' '00108' '00720'
'00450' '00232' '00170' '01160' '00411' '00460' '00348' '00480' '00640'
'00372' '00276' '00221' '00352' '00141' '00178' '00600' '00550' '02000'
'00225' '00210' '00110' '00356' '00045' '00062' '00092' '00174' '00017'
'00086' '00454' '00254' '00028' '00263' '00333' '00312' '00290' '00371'
'00099' '00252' '00760' '00560' '00130' '00523' '00680' '00163' '00208'
'00383' '00330' '00422' '00840' '00432' '00032' '00186' '00303' '00349'
'00224' '00369' '00076' '00231' '00309' '00416' '00465' '00256']
ingreso
[ 0 560 824 3 31285 1349 314 1442 200 2690
245 1208 1260 11 10000 5000 4000 35 713 551
500 300 221 2283 100 15 284 1236 5800 730
400 50000 456 15108 2954 2 20 27 225 1
38 5 130 147 210 11202 1332 50 258 567
1000 2510 809 610 150 51100 367 600 247 375
278 827 2072 582 2300 3065 2200 6 1602 2184
3376 2000 7544 10561 837 11177 639 2028 1065 540
158 15000 3000 3257 1655 1430 7 790 396 678
1187 6590 168 1270 1210 742 8851 7059 1704 857
6700 2503 9800 196 14 26726 18027 99 444 1200
2010 13 120 32 722 40 484 204 98 5552
105 2803 126 4 21 173 10 25 42 100000
113 8 44 2732 179 16 1062 251 228 67
12 122 4208 1300 112 1110 1004 286 4500 1212
195 87 17 184 140 18 146 22 55 70
60 1058 769 5200 19 316 350 3552 687 1950
53 41 33 80 351 2100 475 892 4607 2206
5860 28 1391 2279 591 960 690 234 800 990
2197 90 340 347 327 4071 109 1249 134 1344
321 948 2079 2384 458 5298 162 1583 58 59
1400 1465 8000 4700 1097 3290 13212 5777 5124 23
4159 918 768 283 108 9 68 587 141 501
160 390 154 117 246 237 364 537 394 750]
tarjeta_aprobada
['+' '-']
tc_data.describe()
deudafloat64
años_empleadofloat64
count
690.0
690.0
mean
4.758724637681159
2.223405797101449
std
4.978163248528541
3.3465133592781324
min
0.0
0.0
25%
1.0
0.165
50%
2.75
1.0
75%
7.2075
2.625
max
28.0
28.5
tc_data.describe(include='object')
generoobject
edadobject
count
690
690
unique
2
349
top
b
22.67
freq
480
21
# Eliminando las variables
tc_data = tc_data.drop(["licencia_conduccion", "zip"], axis=1)
tc_data.head()
generoobject
edadobject
0
b
30.83
1
a
58.67
2
a
24.50
3
b
27.83
4
b
20.17
# Instanciar el Label Encoder
le = LabelEncoder()
# Hacer la transformación numérica para cada variable categórica
for col in tc_data.columns.values: # todas las columnas (pasa los nombres como un arreglo)
if tc_data[col].dtypes =='object': # variables categóricas
# Usar el Label Encoder para hacer la transformación numérica
tc_data[col]=le.fit_transform(tc_data[col])
tc_data.head()
generoint64
edadint64
0
1
156
1
0
328
2
0
89
3
1
125
4
1
43
# variables para one-hot encoding
dummy_vars = {"genero": str,
"estado_civil": str,
"cliente": str,
"etnia": str,
"no_pago": str,
"empleado": str,
"ciudadania": str,
"edad": float}
# convierte las variables seleccionadas a categóricas
tc_data = tc_data.astype(dummy_vars)
tc_data.dtypes
# Crear las nuevas variables dummies
tc_data = pd.get_dummies(tc_data, drop_first=True)
tc_data
edadfloat64
0.0 - 348.0
deudafloat64
0.0 - 28.0
0
156.0
0.0
1
328.0
4.46
2
89.0
0.5
3
125.0
1.54
4
43.0
5.625
5
168.0
4.0
6
179.0
1.04
7
74.0
11.585
8
310.0
0.5
9
255.0
4.915
# (Opcional) Re-ordenado las variables: La variable dependiente va al final
# Copiar el target en la última columna
tc_data["ultima"] = tc_data["tarjeta_aprobada"]
# Eliminar la columna inicial del target
tc_data = tc_data.drop("tarjeta_aprobada", axis=1)
# Nombrar nuevamente para facilidad
tc_data.rename(columns={"ultima": "tc_aprobada"}, inplace=True)
tc_data
edadfloat64
0.0 - 348.0
deudafloat64
0.0 - 28.0
0
156.0
0.0
1
328.0
4.46
2
89.0
0.5
3
125.0
1.54
4
43.0
5.625
5
168.0
4.0
6
179.0
1.04
7
74.0
11.585
8
310.0
0.5
9
255.0
4.915
# Convertir dataframe a arreglos de numpy
tc_data = tc_data.values
tc_data
# Separar las variables explicativas de la variable objetivo
X,y = tc_data[:,0:22] , tc_data[:,23]
# Segmentación de datos de entrenamiento y test
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=1)
print(f'X_train tiene estructura: {X_train.shape}')
print(f'X_test tiene estructura: {X_test.shape}')
print(f'\n')
print(f'y_train tiene estructura: {y_train.shape}')
print(f'y_test tiene estructura: {y_test.shape}')
X_train tiene estructura: (462, 22)
X_test tiene estructura: (228, 22)
y_train tiene estructura: (462,)
y_test tiene estructura: (228,)
# Casos NO exitosos (0): Aplicaciones no aprobadas
np.count_nonzero(y)
# Casos exitosos (1): Aplicaciones aprobadas
y.size - np.count_nonzero(y)
# Revisión de rangos de las varibales explicativas
pd.DataFrame(X_train).describe()
0float64
1float64
count
462.0
462.0
mean
145.08658008658008
4.681255411255411
std
93.84343976064402
4.880873662029029
min
1.0
0.0
25%
69.0
1.0
50%
130.0
2.75
75%
216.75
7.0
max
348.0
28.0
# Reescalar X_train y X_test
scaler = MinMaxScaler(feature_range=(0,1)) # instanciar
rescaledX_train = scaler.fit_transform(X_train) # reescalar X de entrenamiento
rescaledX_test = scaler.fit_transform(X_test) # reescalar X de test
# Revisar que los rangos de X_train reescalado sean entre 0 y 1
pd.DataFrame(rescaledX_train).describe()
0float64
1float64
count
462.0
462.0
mean
0.4152351011140636
0.16718769325912183
std
0.2704421895119424
0.17431691650103673
min
0.0
0.0
25%
0.19596541786743515
0.03571428571428571
50%
0.3717579250720461
0.09821428571428571
75%
0.621757925072046
0.25
max
0.9999999999999998
1.0
# Revisa que los rangos de X_test reescalado sean entre 0 y 1
pd.DataFrame(rescaledX_test).describe()
0float64
1float64
count
228.0
228.0
mean
0.4234162495576116
0.18666040457132962
std
0.2637629094314213
0.19658590292441333
min
0.0
0.0
25%
0.20461095100864551
0.03949117144484526
50%
0.3659942363112392
0.11391684070628441
75%
0.6282420749279538
0.2776722992215683
max
0.9999999999999999
1.0
# Instanciar un clasificador de Regresión Logística
logreg = LogisticRegression()
# Ajustar una regresión a los datos de entrenamiento
logreg.fit(rescaledX_train, y_train)
# Usa logreg para predecir aplicaciones de los datos de test
y_pronostico = logreg.predict(rescaledX_test)
# Matriz de confusión para el modelo logreg
matriz_conf = confusion_matrix(y_test, y_pronostico)
plt.figure(figsize=(10, 5))
ax = plt.subplot()
sn.heatmap(matriz_conf, annot=True, ax = ax, cmap='Blues')
plt.xlabel('Predicted')
plt.ylabel('True')
plt.show()
# Evaluar el modelo en los datos de entrenamiento
print("Precisión del clasificador de Reg. logística en train: ", logreg.score(rescaledX_train, y_train))
print("Precisión del clasificador de Reg. logística en test: ", logreg.score(rescaledX_test, y_test))
Precisión del clasificador de Reg. logística en train: 0.8701298701298701
Precisión del clasificador de Reg. logística en test: 0.8640350877192983
from sklearn.model_selection import GridSearchCV
# Parametros
tol = [0.01, 0.001, 0.0001]
max_iter = [150, 200, 300, 1000]
# Crea un diccionario con dos llaves donde tol y max_iter son las llaves y las listas de arriba sus valores correspondientes
param_grid = dict(tol = tol, max_iter = max_iter)
param_grid # imprime para ver el diccionario
# Crear la instancia de GridSearchCV y guárdala como una variable
logreg = LogisticRegression()
grid_model = GridSearchCV(estimator = logreg,
param_grid = param_grid,
cv=5)
# Usar "scaler" para reescalar X y guárdalo en una variable
scaler = MinMaxScaler(feature_range=(0,1))
rescaledX = scaler.fit_transform(X)
# Ajustar los datos a grid_model
grid_model_result = grid_model.fit(rescaledX, y)
# Summarize results
best_score, best_params = grid_model_result.best_score_, grid_model_result.best_params_
# Creating a dictionary to save the best results
best_models = {'Logistic': best_score}
print(f'Mejor resultado: {best_score}, usando {best_params}')
Mejor resultado: 0.8478260869565217, usando {'max_iter': 150, 'tol': 0.01}