머신러닝스터디/2016/2016 09 10 - Revision history

imported>rabierre at 03:39, 13 September 2016

2016-09-13T03:39:52Z

← Older revision		Revision as of 03:39, 13 September 2016
Line 7:		Line 7:
	** Normalization 이용		** Normalization 이용
	*** `X_train/255.` 부분이 Normalization		*** `X_train/255.` 부분이 Normalization
			*** feature의 범위를 0~255에서 0~1 사이의 실수로 scaling 함.
	=== 코드 ===		=== 코드 ===
	import pandas as pd		import pandas as pd

imported>rabierre at 03:38, 13 September 2016

2016-09-13T03:38:37Z

← Older revision		Revision as of 03:38, 13 September 2016
Line 7:		Line 7:
	** Normalization 이용		** Normalization 이용
	*** `X_train/255.` 부분이 Normalization		*** `X_train/255.` 부분이 Normalization
	~~` X_train, X_test, y_train, y_test = train_test_split(X_train/255., y_train, test_size=0.30) `~~
	=== 코드 ===		=== 코드 ===
	import pandas as pd		import pandas as pd

imported>rabierre at 03:38, 13 September 2016

2016-09-13T03:38:16Z

← Older revision		Revision as of 03:38, 13 September 2016
Line 4:		Line 4:
	* Project		* Project
	** [https://www.kaggle.com/c/digit-recognizer kaggle digit recignizer]		** [https://www.kaggle.com/c/digit-recognizer kaggle digit recignizer]
	* 지난 시간([[머신러닝스터디/2016/2016_09_03]]) 학습이 안되는 문제 해결		* 지난 시간( [[머신러닝스터디/2016/2016_09_03]]) 학습이 안되는 문제 해결
	** Normalization 이용		** Normalization 이용
	X_train, X_test, y_train, y_test = train_test_split(X_train/255., y_train, test_size=0.30)		*** `X_train/255.` 부분이 Normalization
			` X_train, X_test, y_train, y_test = train_test_split(X_train/255., y_train, test_size=0.30) `
	=== 코드 ===		=== 코드 ===
	import pandas as pd		import pandas as pd

imported>rabierre: {CREATE}

2016-09-13T03:36:54Z

{CREATE}

New page

[[머신러닝스터디/2016]]
[[머신러닝스터디/2016/목차]]
== 내용 ==
* Project
** [https://www.kaggle.com/c/digit-recognizer kaggle digit recignizer]
* 지난 시간([[머신러닝스터디/2016/2016_09_03]]) 학습이 안되는 문제 해결
** Normalization 이용
X_train, X_test, y_train, y_test = train_test_split(X_train/255., y_train, test_size=0.30)
=== 코드 ===
import pandas as pd
import keras
from sklearn.cross_validation import train_test_split
from keras.utils.np_utils import to_categorical

train = pd.read_csv("../input/train.csv")
test = pd.read_csv("../input/test.csv")

y_train = train['label'].as_matrix()
X_train = train.drop('label', axis=1).as_matrix()

X_train, X_test, y_train, y_test = train_test_split(X_train/255., y_train, test_size=0.30)

model = keras.models.Sequential()

model.add(keras.layers.Dense(64, input_dim=28*28, activation='relu'))
model.add(keras.layers.Dense(32, activation='relu'))
model.add(keras.layers.Dense(16, activation='relu'))
model.add(keras.layers.Dense(10, activation='softmax'))

model.compile(loss='categorical_crossentropy', optimizer='adagrad', metrics=['accuracy'])
model.fit(X_train, to_categorical(y_train, 10), nb_epoch=5, batch_size=600)

score = model.evaluate(X_test, to_categorical(y_test, 10), batch_size=700)

print(score)

print(model.predict(X_test))[0]
print(y_test[0])
== 다음 시간에는 ==
* 한 주 쉬어갑니다.