241031

241031

1. 회귀

In [1]:

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.metrics import *
from sklearn.preprocessing import MinMaxScaler

from keras.models import Sequential
from keras.layers import Dense, Input
from keras.backend import clear_session
from keras.optimizers import Adam

In [2]:

path = 'https://raw.githubusercontent.com/DA4BAM/dataset/master/Carseats.csv'
data = pd.read_csv(path)
data

Out[2]:

SalesCompPriceIncomeAdvertisingPopulationPriceShelveLocAgeEducationUrbanUS01234...395396397398399

9.50	138	73	11	276	120	Bad	42	17	Yes	Yes
11.22	111	48	16	260	83	Good	65	10	Yes	Yes
10.06	113	35	10	269	80	Medium	59	12	Yes	Yes
7.40	117	100	4	466	97	Medium	55	14	Yes	Yes
4.15	141	64	3	340	128	Bad	38	13	Yes	No
...	...	...	...	...	...	...	...	...	...	...
12.57	138	108	17	203	128	Good	33	14	Yes	Yes
6.14	139	23	3	37	120	Medium	55	11	No	Yes
7.41	162	26	12	368	159	Medium	40	18	Yes	Yes
5.94	100	79	7	284	95	Bad	50	12	Yes	Yes
9.71	134	37	0	27	120	Good	49	16	Yes	Yes

400 rows × 11 columns

 

In [3]:

# 가변수화
x = data.drop(columns = 'Sales')
y = data['Sales']

x = pd.get_dummies(x, columns = ['ShelveLoc', 'Education', 'US', 'Urban'], drop_first = True)

In [4]:

# 데이터 분할
x_train, x_val, y_train, y_val = train_test_split(x, y, random_state = 1, test_size = 0.3)

# 스케일링
scaler = MinMaxScaler()
x_train = scaler.fit_transform(x_train)
x_val = scaler.transform(x_val)

In [5]:

# 모델링
nfeatures = x_train.shape[1]

clear_session()

model1 = Sequential([Input(shape = (nfeatures, )),
                     Dense(2, activation = 'relu'),
                     Dense(1)])
model1.summary()

 

 

In [6]:

# 컴파일, 학습
model1.compile(optimizer = Adam(learning_rate = 0.01), loss = 'mse')
hist1 = model1.fit(x_train, y_train, epochs = 500, validation_split = 0.3, verbose = 0).history

In [7]:

plt.plot(hist1['loss'], label = 'train_err', marker = '.')
plt.plot(hist1['val_loss'], label= 'val_err', marker = '.')
plt.grid()
plt.legend()
plt.show()

In [8]:

pred1 = model1.predict(x_val)
print(mean_absolute_error(y_val, pred1))

4/4 ━━━━━━━━━━━━━━━━━━━━ 0s 11ms/step
0.9141374322573343

2. 이진분류

 

In [9]:

path = "https://raw.githubusercontent.com/DA4BAM/dataset/master/Attrition_train_validation.CSV"
data = pd.read_csv(path)

# 숫자형 데이터로 변경
data['Attrition'] = np.where(data['Attrition']=='Yes', 1, 0)
data.tail()

Out[9]:

AttritionAgeBusinessTravelDepartmentDistanceFromHomeEducationEducationFieldEmployeeNumberEnvironmentSatisfactionGender...OverTimePercentSalaryHikeRelationshipSatisfactionStockOptionLevelTotalWorkingYearsTrainingTimesLastYearWorkLifeBalanceYearsAtCompanyYearsInCurrentRoleYearsWithCurrManager12451246124712481249

0	27	Travel_Rarely	Research & Development	19	3	Other	1619	4	Male	...	No	11	1	2	7	3	3	7	7	7
0	29	Travel_Rarely	Research & Development	9	3	Life Sciences	1558	3	Male	...	No	18	1	2	5	2	2	1	0	0
0	29	Non-Travel	Sales	2	3	Life Sciences	469	4	Male	...	No	14	1	1	4	3	2	4	3	2
0	43	Travel_Rarely	Sales	16	3	Marketing	327	4	Female	...	Yes	22	3	1	22	3	3	17	13	9
0	26	Travel_Rarely	Research & Development	3	4	Medical	1177	1	Male	...	No	22	2	1	8	2	3	8	7	7

5 rows × 26 columns

 

In [10]:

# 불필요한 변수 제거
data.drop(columns = 'EmployeeNumber', inplace = True)

x = data.drop(columns = 'Attrition')
y = data['Attrition']

# 가변수화
dum_cols = ['BusinessTravel','Department','Education','EducationField','EnvironmentSatisfaction','Gender',
            'JobRole', 'JobInvolvement', 'JobSatisfaction', 'MaritalStatus', 'OverTime', 'RelationshipSatisfaction',
            'StockOptionLevel','WorkLifeBalance' ]
x = pd.get_dummies(x, columns = dum_cols ,drop_first = True)

In [11]:

# 데이터 분할
x_train, x_val, y_train, y_val = train_test_split(x, y, test_size = 0.3, random_state = 1)

# 스케일링
scaler = MinMaxScaler()
x_train = scaler.fit_transform(x_train)
x_val = scaler.transform(x_val)

In [12]:

nfeatures = x_train.shape[1]
nfeatures

Out[12]:

53

In [13]:

# 모델링
clear_session()

model2 = Sequential([Input(shape = (nfeatures, )),
                     Dense(8, activation = 'relu'),
                     Dense(4, activation = 'relu'),
                     Dense(1, activation = 'sigmoid')])

model2.summary()

 

 

In [14]:

# 컴파일, 학습
model2.compile(optimizer = Adam(learning_rate = 0.001), loss = 'binary_crossentropy')
hist2 = model2.fit(x_train, y_train, epochs = 50, validation_split = 0.2, verbose = 0).history

In [15]:

plt.plot(hist2['loss'], label = 'train_err', marker = '.')
plt.plot(hist2['val_loss'], label= 'val_err', marker = '.')
plt.grid()
plt.legend()
plt.show()

In [16]:

# y값 전처리
pred2 = model2.predict(x_val)
pred2 = np.where(pred2> 0.5, 1, 0)

print(classification_report(y_val, pred2))

12/12 ━━━━━━━━━━━━━━━━━━━━ 0s 4ms/step 
              precision    recall  f1-score   support

           0       0.87      0.98      0.92       315
           1       0.67      0.23      0.35        60

    accuracy                           0.86       375
   macro avg       0.77      0.61      0.63       375
weighted avg       0.84      0.86      0.83       375

3. 다중분류

In [17]:

path = "https://raw.githubusercontent.com/DA4BAM/dataset/master/iris.csv"
data = pd.read_csv(path)
data['Species'] = data['Species'].map({'setosa':0, 'versicolor':1, 'virginica':2})
data.head()

Out[17]:

Sepal.LengthSepal.WidthPetal.LengthPetal.WidthSpecies01234

5.1	3.5	1.4	0.2	0
4.9	3.0	1.4	0.2	0
4.7	3.2	1.3	0.2	0
4.6	3.1	1.5	0.2	0
5.0	3.6	1.4	0.2	0

 

In [18]:

# 데이터 분할
x = data.drop(columns = 'Species')
y = data['Species']
x_train, x_val, y_train, y_val = train_test_split(x, y, test_size = 0.3, random_state = 1)

# 스케일
scaler = MinMaxScaler()
x_train = scaler.fit_transform(x_train)
x_val = scaler.transform(x_val)

In [19]:

nfeatures = x_train.shape[1]
nfeatures

Out[19]:

4

In [25]:

# 모델링
clear_session()

model3 = Sequential([Input(shape = (nfeatures, )),
                     Dense(3, activation = 'softmax')])

model3.summary()

 

 

In [26]:

# 컴파일, 학습
model3.compile(optimizer = Adam(learning_rate = 0.1), loss = 'sparse_categorical_crossentropy')
hist3 = model3.fit(x_train, y_train, validation_split = 0.3, epochs = 50, verbose = 0).history

In [27]:

plt.plot(hist3['loss'], label = 'train_err', marker = '.')
plt.plot(hist3['val_loss'], label= 'val_err', marker = '.')
plt.grid()
plt.legend()
plt.show()

In [30]:

# y값 전처리
pred3 = model3.predict(x_val)
pred3 = pred3.argmax(axis = 1)

print(classification_report(y_val, pred3))

2/2 ━━━━━━━━━━━━━━━━━━━━ 0s 7ms/step 
              precision    recall  f1-score   support

           0       1.00      1.00      1.00        14
           1       0.94      0.89      0.91        18
           2       0.86      0.92      0.89        13

    accuracy                           0.93        45
   macro avg       0.93      0.94      0.93        45
weighted avg       0.94      0.93      0.93        45

In [ ]: