在新数据集中进行预测

Question

I build a keras logistic regression model. I am trying to find a way that i could give my model new data-set and give me prediction in the new data set that i passed. my new data-set will be the same shape of my model

my second question is there a way to improve the accuracy of my model becouse my accrucy is 69% and when i print the classification repoert i got bad precion in one class

X=new.drop('reassed',axis=1)
y=new['reassed'].astype(int)

split data

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 0)


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

# Initialising the ANN
classifier = Sequential()

# Adding the input layer and the first hidden layer
classifier.add(Dense(units = 27, kernel_initializer = 'uniform', activation = 'relu', input_dim = 6))

# Adding the second hidden layer
classifier.add(Dense(units = 27, kernel_initializer = 'uniform', activation = 'relu'))

# Adding the output layer
classifier.add(Dense(units = 1, kernel_initializer = 'uniform', activation = 'sigmoid'))

# Compiling the ANN
classifier.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])`enter code here`

# Fitting the ANN to the Training set
classifier.fit(X_train, y_train, batch_size = 10, epochs = 20)


Epoch 1/20
16704/16704 [==============================] - 1s 76us/step - loss: 0.6159 - acc: 0.6959
Epoch 2/20
16704/16704 [==============================] - 1s 65us/step - loss: 0.6114 - acc: 0.6967
Epoch 3/20
16704/16704 [==============================] - 1s 65us/step - loss: 0.6110 - acc: 0.6964
Epoch 4/20
16704/16704 [==============================] - 1s 66us/step - loss: 0.6101 - acc: 0.6965
Epoch 5/20
16704/16704 [==============================] - 1s 66us/step - loss: 0.6091 - acc: 0.6961
Epoch 6/20
16704/16704 [==============================] - 1s 66us/step - loss: 0.6094 - acc: 0.6963
Epoch 7/20
16704/16704 [==============================] - 1s 68us/step - loss: 0.6086 - acc: 0.6967
Epoch 8/20
16704/16704 [==============================] - 1s 66us/step - loss: 0.6083 - acc: 0.6965
Epoch 9/20
16704/16704 [==============================] - 1s 65us/step - loss: 0.6081 - acc: 0.6964: 0s - loss: 0.6085 - acc: 
Epoch 10/20
16704/16704 [==============================] - 1s 66us/step - loss: 0.6082 - acc: 0.6971
Epoch 11/20
16704/16704 [==============================] - 1s 67us/step - loss: 0.6077 - acc: 0.6968
Epoch 12/20
16704/16704 [==============================] - 1s 66us/step - loss: 0.6073 - acc: 0.6971
Epoch 13/20
16704/16704 [==============================] - 1s 65us/step - loss: 0.6067 - acc: 0.6971
Epoch 14/20
16704/16704 [==============================] - 1s 66us/step - loss: 0.6070 - acc: 0.6965
Epoch 15/20
16704/16704 [==============================] - 1s 65us/step - loss: 0.6066 - acc: 0.6967: 0s - loss: 0.6053 - ac
Epoch 16/20
16704/16704 [==============================] - 1s 66us/step - loss: 0.6060 - acc: 0.6967
Epoch 17/20
16704/16704 [==============================] - 1s 67us/step - loss: 0.6061 - acc: 0.6968
Epoch 18/20
16704/16704 [==============================] - 1s 67us/step - loss: 0.6062 - acc: 0.6971
Epoch 19/20
16704/16704 [==============================] - 1s 69us/step - loss: 0.6057 - acc: 0.6968
Epoch 20/20
16704/16704 [==============================] - 1s 74us/step - loss: 0.6055 - acc: 0.6973

y_pred = classifier.predict(X_test)
y_pred = [ 1 if y>=0.5 else 0 for y in y_pred ]

print(classification_report(y_test, y_pred))

      precision    recall  f1-score   support

           0       0.71      1.00      0.83      2968
           1       0.33      0.00      0.01      1208

   micro avg       0.71      0.71      0.71      4176
   macro avg       0.52      0.50      0.42      4176
weighted avg       0.60      0.71      0.59      4176

I expect to improve my model

I expect to find a way that i could make prediction in new data-set

Answer 1

To make prediction on the new data set

1. Load the data the same you load your test set
2. Apply all the per-processing steps applied on your training set.

3. Use the

   model.predict(X)

function to make prediction and carry on with your post processing.

It's almost same as predicting with the test set.