In [1]:
# Tratamiento de datos
# ==============================================================================
import pandas as pd
import numpy as np
# Preprocesado y modelado
# ==============================================================================
from sklearn.svm import SVC
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
# Configuración warnings
# ==============================================================================
import warnings
warnings.filterwarnings('ignore')
In [2]:
#Cargamos los datos!
url = 'https://raw.githubusercontent.com/JoaquinAmatRodrigo/' \
+ 'Estadistica-machine-learning-python/master/data/ESL.mixture.csv'
datos = pd.read_csv(url)
datos.head()
Out[2]:
| X1 | X2 | y | |
|---|---|---|---|
| 0 | 2.526093 | 0.321050 | 0 |
| 1 | 0.366954 | 0.031462 | 0 |
| 2 | 0.768219 | 0.717486 | 0 |
| 3 | 0.693436 | 0.777194 | 0 |
| 4 | -0.019837 | 0.867254 | 0 |
In [3]:
datos.y.unique()
Out[3]:
array([0, 1])
In [4]:
datos.shape
Out[4]:
(200, 3)
In [5]:
datos.y.unique()
Out[5]:
array([0, 1])
In [6]:
#Visualizacion!
import matplotlib.pyplot as plt
fig, ax = plt.subplots(figsize=(6,4))
ax.scatter(datos.X1, datos.X2, c=datos.y);
ax.set_title("Datos");
In [7]:
# División de los datos en train y test
X = datos.drop(columns = 'y') # Matriz de 2 columnas -Variable Independientes
y = datos['y'] # Vector de 200 filas y 1 columa- Variable Dependiente
In [8]:
X
Out[8]:
| X1 | X2 | |
|---|---|---|
| 0 | 2.526093 | 0.321050 |
| 1 | 0.366954 | 0.031462 |
| 2 | 0.768219 | 0.717486 |
| 3 | 0.693436 | 0.777194 |
| 4 | -0.019837 | 0.867254 |
| ... | ... | ... |
| 195 | 0.256750 | 2.293605 |
| 196 | 1.925173 | 0.165053 |
| 197 | 1.301941 | 0.992200 |
| 198 | 0.008131 | 2.242264 |
| 199 | -0.196246 | 0.551404 |
200 rows × 2 columns
In [9]:
y
Out[9]:
0 0
1 0
2 0
3 0
4 0
..
195 1
196 1
197 1
198 1
199 1
Name: y, Length: 200, dtype: int64
In [10]:
y.values.reshape(-1,1)
Out[10]:
array([[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[0],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1],
[1]])
In [11]:
X_train, X_test, y_train, y_test = train_test_split(X,y.values.reshape(-1,1),train_size= 0.75,\
random_state = 42,shuffle=True)
gamma es un parámetro para hiperplanos no lineales. Cuanto mayor sea el valor de gamma, intentará ajustarse exactamente al conjunto de datos de entrenamiento.
C es el parámetro de penalización del término de error. Controla el equilibrio entre un límite de decisión uniforme y la clasificación correcta de los puntos de entrenamiento.
In [19]:
SVC?
In [20]:
# Creación del modelo SVM
modelo = SVC(C=0.1,kernel='rbf',random_state=42)
modelo.fit(X_train, y_train)
Out[20]:
SVC(C=0.1, random_state=42)
In [21]:
#Predicciones!
#y_train_pred = modelo.predict(X_train)
y_test_pred = modelo.predict(X_test)
In [22]:
X_test
Out[22]:
| X1 | X2 | |
|---|---|---|
| 95 | 0.434501 | -0.064545 |
| 15 | 1.279560 | -0.271033 |
| 30 | 2.988126 | 0.672351 |
| 158 | 0.508118 | 1.757232 |
| 128 | -1.136539 | 2.052686 |
| 115 | -2.017571 | 2.720377 |
| 69 | 2.482942 | -0.737176 |
| 170 | 0.571872 | 1.577942 |
| 174 | 1.244438 | 0.852323 |
| 45 | 1.749188 | -0.357916 |
| 66 | 2.290042 | 0.137390 |
| 182 | 1.518294 | 0.361734 |
| 165 | 3.503126 | 1.223982 |
| 78 | 0.087615 | -0.733347 |
| 186 | 1.486248 | 0.918180 |
| 177 | 1.287422 | 0.272585 |
| 56 | 0.134347 | 0.901676 |
| 152 | -1.960780 | 2.002274 |
| 82 | 2.576843 | -1.435682 |
| 68 | 0.516429 | 0.525152 |
| 124 | -1.464691 | 2.473922 |
| 16 | -0.313007 | 1.273747 |
| 148 | -2.477418 | 1.026506 |
| 93 | 1.419537 | -0.029746 |
| 65 | 0.310785 | 2.007982 |
| 60 | 0.325769 | 0.347996 |
| 84 | 2.892025 | 1.625783 |
| 67 | 0.659906 | -1.241637 |
| 125 | 1.623802 | 0.240937 |
| 132 | -0.071295 | 1.460543 |
| 9 | 1.897709 | 0.973755 |
| 18 | 0.319988 | 0.017202 |
| 55 | -0.727077 | 1.457361 |
| 75 | 0.951265 | -1.095890 |
| 150 | 1.028464 | 1.550986 |
| 104 | 1.076957 | -0.330735 |
| 135 | 0.433583 | 1.261365 |
| 137 | 1.432136 | 0.306475 |
| 164 | -0.570430 | 1.090314 |
| 76 | 0.425476 | 0.184899 |
| 79 | 2.285265 | 1.008993 |
| 197 | 1.301941 | 0.992200 |
| 38 | 1.532258 | -1.082164 |
| 24 | 1.467109 | 1.877591 |
| 122 | 1.334411 | -0.294441 |
| 195 | 0.256750 | 2.293605 |
| 29 | 2.126869 | 1.154180 |
| 19 | 0.437542 | 0.413648 |
| 143 | 1.047458 | 1.039996 |
| 86 | -0.074079 | 0.919702 |
In [23]:
X_test.shape
Out[23]:
(50, 2)
In [24]:
y_test_pred
Out[24]:
array([0, 0, 0, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 1, 1,
1, 0, 1, 0, 1, 0, 0, 1, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 1, 0, 1,
0, 1, 1, 0, 1, 1])
In [25]:
y_test
Out[25]:
array([[0],
[0],
[0],
[1],
[1],
[1],
[0],
[1],
[1],
[0],
[0],
[1],
[1],
[0],
[1],
[1],
[0],
[1],
[0],
[0],
[1],
[0],
[1],
[0],
[0],
[0],
[0],
[0],
[1],
[1],
[0],
[0],
[0],
[0],
[1],
[1],
[1],
[1],
[1],
[0],
[0],
[1],
[0],
[0],
[1],
[1],
[0],
[0],
[1],
[0]])
In [26]:
#Calculo el accuracy en Test
from sklearn.metrics import accuracy_score
test_accuracy = accuracy_score(y_test, y_test_pred)
print('% de aciertos sobre el set de evaluación:',test_accuracy)
% de aciertos sobre el set de evaluación: 0.68
Created in