12장 다중 분류 문제 해결하기¶

2. 상관도 그래프¶

In [1]:

import pandas as pd

# 깃허브에 준비된 데이터를 가져옵니다.
!git clone https://github.com/taehojo/data.git

# 아이리스 데이터를 불러옵니다.
df = pd.read_csv('./data/iris3.csv')

import pandas as pd # 깃허브에 준비된 데이터를 가져옵니다. !git clone https://github.com/taehojo/data.git # 아이리스 데이터를 불러옵니다. df = pd.read_csv('./data/iris3.csv')

In [2]:

# 첫 5줄을 봅니다.
df.head()

# 첫 5줄을 봅니다. df.head()

Out[2]:

	sepal_length	sepal_width	petal_length	petal_width	species
0	5.1	3.5	1.4	0.2	Iris-setosa
1	4.9	3.0	1.4	0.2	Iris-setosa
2	4.7	3.2	1.3	0.2	Iris-setosa
3	4.6	3.1	1.5	0.2	Iris-setosa
4	5.0	3.6	1.4	0.2	Iris-setosa

In [3]:

import seaborn as sns
import matplotlib.pyplot as plt

# 그래프로 확인해 봅시다.
sns.pairplot(df, hue='species');
plt.show()

import seaborn as sns import matplotlib.pyplot as plt # 그래프로 확인해 봅시다. sns.pairplot(df, hue='species'); plt.show()

3. 원-핫 인코딩¶

In [4]:

# 속성을 X, 클래스를 y로 저장합니다.
X = df.iloc[:,0:4]
y = df.iloc[:,4]

# X와 y의 첫 5줄을 출력해 보겠습니다. 
print(X[0:5])
print(y[0:5])

# 속성을 X, 클래스를 y로 저장합니다. X = df.iloc[:,0:4] y = df.iloc[:,4] # X와 y의 첫 5줄을 출력해 보겠습니다. print(X[0:5]) print(y[0:5])

   sepal_length  sepal_width  petal_length  petal_width
0           5.1          3.5           1.4          0.2
1           4.9          3.0           1.4          0.2
2           4.7          3.2           1.3          0.2
3           4.6          3.1           1.5          0.2
4           5.0          3.6           1.4          0.2
0    Iris-setosa
1    Iris-setosa
2    Iris-setosa
3    Iris-setosa
4    Iris-setosa
Name: species, dtype: object

In [5]:

# 원-핫 인코딩 처리를 합니다.
y = pd.get_dummies(y)

# 원-핫 인코딩 결과를 확인합니다.
print(y[0:5])

# 원-핫 인코딩 처리를 합니다. y = pd.get_dummies(y) # 원-핫 인코딩 결과를 확인합니다. print(y[0:5])

   Iris-setosa  Iris-versicolor  Iris-virginica
0            1                0               0
1            1                0               0
2            1                0               0
3            1                0               0
4            1                0               0

4. 소프트맥스¶

In [11]:

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

# 모델 설정
model = Sequential()
model.add(Dense(12,  input_dim=4, activation='relu'))
model.add(Dense(8,  activation='relu'))
model.add(Dense(3, activation='softmax'))
model.summary()

# 모델 컴파일
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

# 모델 실행
history=model.fit(X, y, epochs=30, batch_size=5)

from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense # 모델 설정 model = Sequential() model.add(Dense(12, input_dim=4, activation='relu')) model.add(Dense(8, activation='relu')) model.add(Dense(3, activation='softmax')) model.summary() # 모델 컴파일 model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy']) # 모델 실행 history=model.fit(X, y, epochs=30, batch_size=5)

Model: "sequential_5"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense_15 (Dense)             (None, 12)                60        
_________________________________________________________________
dense_16 (Dense)             (None, 8)                 104       
_________________________________________________________________
dense_17 (Dense)             (None, 3)                 27        
=================================================================
Total params: 191
Trainable params: 191
Non-trainable params: 0
_________________________________________________________________
Epoch 1/30
30/30 [==============================] - 0s 2ms/step - loss: 1.9954 - accuracy: 0.3333
Epoch 2/30
30/30 [==============================] - 0s 2ms/step - loss: 1.4589 - accuracy: 0.3333
Epoch 3/30
30/30 [==============================] - 0s 2ms/step - loss: 1.2054 - accuracy: 0.3333
Epoch 4/30
30/30 [==============================] - 0s 2ms/step - loss: 1.0958 - accuracy: 0.3533
Epoch 5/30
30/30 [==============================] - 0s 2ms/step - loss: 1.0327 - accuracy: 0.6333
Epoch 6/30
30/30 [==============================] - 0s 2ms/step - loss: 0.9898 - accuracy: 0.6067
Epoch 7/30
30/30 [==============================] - 0s 2ms/step - loss: 0.9552 - accuracy: 0.6000
Epoch 8/30
30/30 [==============================] - 0s 2ms/step - loss: 0.9176 - accuracy: 0.5733
Epoch 9/30
30/30 [==============================] - 0s 2ms/step - loss: 0.8828 - accuracy: 0.6200
Epoch 10/30
30/30 [==============================] - 0s 2ms/step - loss: 0.8463 - accuracy: 0.6200
Epoch 11/30
30/30 [==============================] - 0s 2ms/step - loss: 0.7967 - accuracy: 0.9067
Epoch 12/30
30/30 [==============================] - 0s 2ms/step - loss: 0.7506 - accuracy: 0.9667
Epoch 13/30
30/30 [==============================] - 0s 2ms/step - loss: 0.6934 - accuracy: 0.9600
Epoch 14/30
30/30 [==============================] - 0s 2ms/step - loss: 0.6377 - accuracy: 0.9667
Epoch 15/30
30/30 [==============================] - 0s 2ms/step - loss: 0.5785 - accuracy: 0.9667
Epoch 16/30
30/30 [==============================] - 0s 2ms/step - loss: 0.5282 - accuracy: 0.9400
Epoch 17/30
30/30 [==============================] - 0s 2ms/step - loss: 0.4862 - accuracy: 0.9467
Epoch 18/30
30/30 [==============================] - 0s 2ms/step - loss: 0.4334 - accuracy: 0.9600
Epoch 19/30
30/30 [==============================] - 0s 2ms/step - loss: 0.3968 - accuracy: 0.9667
Epoch 20/30
30/30 [==============================] - 0s 2ms/step - loss: 0.3657 - accuracy: 0.9600
Epoch 21/30
30/30 [==============================] - 0s 2ms/step - loss: 0.3304 - accuracy: 0.9600
Epoch 22/30
30/30 [==============================] - 0s 2ms/step - loss: 0.3052 - accuracy: 0.9733
Epoch 23/30
30/30 [==============================] - 0s 2ms/step - loss: 0.2843 - accuracy: 0.9667
Epoch 24/30
30/30 [==============================] - 0s 2ms/step - loss: 0.2698 - accuracy: 0.9733
Epoch 25/30
30/30 [==============================] - 0s 2ms/step - loss: 0.2493 - accuracy: 0.9600
Epoch 26/30
30/30 [==============================] - 0s 2ms/step - loss: 0.2385 - accuracy: 0.9667
Epoch 27/30
30/30 [==============================] - 0s 2ms/step - loss: 0.2211 - accuracy: 0.9733
Epoch 28/30
30/30 [==============================] - 0s 2ms/step - loss: 0.2104 - accuracy: 0.9733
Epoch 29/30
30/30 [==============================] - 0s 2ms/step - loss: 0.2068 - accuracy: 0.9733
Epoch 30/30
30/30 [==============================] - 0s 2ms/step - loss: 0.1930 - accuracy: 0.9667

5. 아이리스 품족 예측 실행¶

In [10]:

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# 깃허브에 준비된 데이터를 가져옵니다. 앞에서 이미 가져왔으므로 주석 처리합니다. 5번 예제만 별도 실행 시 주석을 해제한 후 실행해주세요.
# !git clone https://github.com/taehojo/data.git

# 아이리스 데이터를 불러옵니다.
df = pd.read_csv('./data/iris3.csv')

# 속성을 X, 클래스를 y로 저장합니다.
X = df.iloc[:,0:4]
y = df.iloc[:,4]

# 원-핫 인코딩 처리를 합니다.
y = pd.get_dummies(y)

# 모델 설정
model = Sequential()
model.add(Dense(12,  input_dim=4, activation='relu'))
model.add(Dense(8,  activation='relu'))
model.add(Dense(3, activation='softmax'))
model.summary()

# 모델 컴파일
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

# 모델 실행
history=model.fit(X, y, epochs=30, batch_size=5)

from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense import pandas as pd import seaborn as sns import matplotlib.pyplot as plt # 깃허브에 준비된 데이터를 가져옵니다. 앞에서 이미 가져왔으므로 주석 처리합니다. 5번 예제만 별도 실행 시 주석을 해제한 후 실행해주세요. # !git clone https://github.com/taehojo/data.git # 아이리스 데이터를 불러옵니다. df = pd.read_csv('./data/iris3.csv') # 속성을 X, 클래스를 y로 저장합니다. X = df.iloc[:,0:4] y = df.iloc[:,4] # 원-핫 인코딩 처리를 합니다. y = pd.get_dummies(y) # 모델 설정 model = Sequential() model.add(Dense(12, input_dim=4, activation='relu')) model.add(Dense(8, activation='relu')) model.add(Dense(3, activation='softmax')) model.summary() # 모델 컴파일 model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy']) # 모델 실행 history=model.fit(X, y, epochs=30, batch_size=5)

Model: "sequential_4"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense_12 (Dense)             (None, 12)                60        
_________________________________________________________________
dense_13 (Dense)             (None, 8)                 104       
_________________________________________________________________
dense_14 (Dense)             (None, 3)                 27        
=================================================================
Total params: 191
Trainable params: 191
Non-trainable params: 0
_________________________________________________________________
Epoch 1/30
30/30 [==============================] - 0s 2ms/step - loss: 1.4888 - accuracy: 0.3333
Epoch 2/30
30/30 [==============================] - 0s 2ms/step - loss: 1.0854 - accuracy: 0.4067
Epoch 3/30
30/30 [==============================] - 0s 2ms/step - loss: 0.9764 - accuracy: 0.6000
Epoch 4/30
30/30 [==============================] - 0s 2ms/step - loss: 0.9382 - accuracy: 0.6267
Epoch 5/30
30/30 [==============================] - 0s 2ms/step - loss: 0.9109 - accuracy: 0.6267
Epoch 6/30
30/30 [==============================] - 0s 2ms/step - loss: 0.8772 - accuracy: 0.6333
Epoch 7/30
30/30 [==============================] - 0s 2ms/step - loss: 0.8486 - accuracy: 0.6133
Epoch 8/30
30/30 [==============================] - 0s 2ms/step - loss: 0.8224 - accuracy: 0.6267
Epoch 9/30
30/30 [==============================] - 0s 2ms/step - loss: 0.7977 - accuracy: 0.6333
Epoch 10/30
30/30 [==============================] - 0s 2ms/step - loss: 0.7711 - accuracy: 0.6333
Epoch 11/30
30/30 [==============================] - 0s 2ms/step - loss: 0.7469 - accuracy: 0.6200
Epoch 12/30
30/30 [==============================] - 0s 2ms/step - loss: 0.7271 - accuracy: 0.6267
Epoch 13/30
30/30 [==============================] - 0s 2ms/step - loss: 0.7038 - accuracy: 0.6333
Epoch 14/30
30/30 [==============================] - 0s 2ms/step - loss: 0.6802 - accuracy: 0.6333
Epoch 15/30
30/30 [==============================] - 0s 2ms/step - loss: 0.6672 - accuracy: 0.6400
Epoch 16/30
30/30 [==============================] - 0s 3ms/step - loss: 0.6392 - accuracy: 0.6333
Epoch 17/30
30/30 [==============================] - 0s 3ms/step - loss: 0.6284 - accuracy: 0.6200
Epoch 18/30
30/30 [==============================] - 0s 2ms/step - loss: 0.5992 - accuracy: 0.6333
Epoch 19/30
30/30 [==============================] - 0s 2ms/step - loss: 0.5717 - accuracy: 0.6400
Epoch 20/30
30/30 [==============================] - 0s 2ms/step - loss: 0.5328 - accuracy: 0.6933
Epoch 21/30
30/30 [==============================] - 0s 2ms/step - loss: 0.4983 - accuracy: 0.9200
Epoch 22/30
30/30 [==============================] - 0s 2ms/step - loss: 0.4664 - accuracy: 0.9667
Epoch 23/30
30/30 [==============================] - 0s 2ms/step - loss: 0.4439 - accuracy: 0.9533
Epoch 24/30
30/30 [==============================] - 0s 2ms/step - loss: 0.4112 - accuracy: 0.9667
Epoch 25/30
30/30 [==============================] - 0s 2ms/step - loss: 0.3791 - accuracy: 0.9733
Epoch 26/30
30/30 [==============================] - 0s 2ms/step - loss: 0.3551 - accuracy: 0.9667
Epoch 27/30
30/30 [==============================] - 0s 2ms/step - loss: 0.3310 - accuracy: 0.9800
Epoch 28/30
30/30 [==============================] - 0s 2ms/step - loss: 0.3149 - accuracy: 0.9667
Epoch 29/30
30/30 [==============================] - 0s 2ms/step - loss: 0.2894 - accuracy: 0.9733
Epoch 30/30
30/30 [==============================] - 0s 2ms/step - loss: 0.2746 - accuracy: 0.9667

In [ ]: