如何做出新的預測

Question

讓我解釋一下，我正在使用人工神經網絡。 該模型具有15個變量，14個獨立變量和一個相關變量。 在自變量中，我有3個類別變量(day of week, month, direction(north,south, etc...)) 。 我已經對它們進行編碼(monday = 1, tuesday = 2, and so...),我也對它們進行了熱編碼(monday = [1,0,0,0], tuesday = [0,1,0,0]) 。

我的問題是，我該如何用新的值進行預測，就像這樣。

X=['Monday','January','South']

這是代碼

# Classification template

# Importing the libraries
import numpy as np
import pandas as pd

# Importing the dataset

dataset = pd.read_csv('clean.csv')

X = dataset.iloc[:, [4,5,6,9,12,15,16]].values
y = dataset.iloc[:, 14].values

#Encoding categorical Data
from sklearn.preprocessing import LabelEncoder, OneHotEncoder
labelenconder_X = LabelEncoder()
X[:,1] = labelenconder_X.fit_transform(X[:,1])

labelenconder_X_2 = LabelEncoder()
X[:,2] = labelenconder_X_2.fit_transform(X[:,2])

labelenconder_X_7 = LabelEncoder()
X[:,4] = labelenconder_X_7.fit_transform(X[:,4])

labelenconder_X_9 = LabelEncoder()
X[:,5] = labelenconder_X_9.fit_transform(X[:,5])

labelenconder_X_10 = LabelEncoder()
X[:,6] = labelenconder_X_10.fit_transform(X[:,6])

onehotencoder = OneHotEncoder(categorical_features=[1,2,4,5,6])

X = onehotencoder.fit_transform(X).toarray()

X = X[:, 1:]



# Splitting the dataset into the Training set and Test set
from sklearn.cross_validation import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.25, random_state = 0)

# Feature Scaling
#from sklearn.preprocessing import StandardScaler
#sc = StandardScaler()
#X_train = sc.fit_transform(X_train)
#X_test = sc.transform(X_test)

# Fitting classifier to the Training set
# Create your classifier here
import keras
from keras.models import Sequential
from keras.layers import Dense


classifier = Sequential()

                    #INPUT LAYER AND HIDDEN LAYER
classifier.add(Dense(units = 5, kernel_initializer = 'uniform', activation = 'relu', input_dim =9))

                    #ADDING SECOND HIDDEN LAYER
classifier.add(Dense(units = 5, kernel_initializer = 'uniform', activation =  'relu'))

                    #adding output node 
classifier.add(Dense(units= 1, kernel_initializer = 'uniform', activation = 'sigmoid'))

                    #Applygin Stochasting Gradient Descent

classifier.compile(optimizer='adam', loss = 'binary_crossentropy', metrics=['accuracy'])

classifier.fit(X_train, y_train, batch_size =28, epochs = 100)


classifier.save('ANN2.h5')
model = keras.models.load_model('ANN2.h5')
y_predict = model.predict(X_test)
y_predict = (y_predict > 0.40)

Answer 1

如果要對一周中的所有天進行編碼以進行預測，則monday可能應為[1,0,0,0,0,0,0] 。 或者您使用回歸（0.0-6.0）而不是分類。

但是，由於您在這里使用X而不是y ，所以我不確定您提供的X=['Monday','January','South']是輸入而不是輸出（ y ）。 如果是這樣，則不需要單次編碼，而可以將其編碼為例如X=[0,0,2]

• 星期一0
• 星期二1
• ...
• 一月0
• 2月1日
• 北0
• 東1
• ...

我同意@morsecodist的觀點，即需要更多信息才能正確回答您的問題。