如何将model.predict的结果返回到我的pandas DataFrame中

Question

下面可以看到我的模型的实现。

feature_names_activity = ['org:resource', 'lifecycletransition', 'case:concept:name',
'case:AMOUNT_REQ', 'Duration', 'day', 'hour', 'prev_activity', 'position', 
'A_ACCEPTED', 'A_ACTIVATED', 'A_APPROVED', 'A_CANCELLED', 'A_DECLINED', 'A_FINALIZED', 
'A_PARTLYSUBMITTED', 'A_PREACCEPTED', 'A_REGISTERED', 'A_SUBMITTED', 'O_ACCEPTED', 'O_CANCELLED',
'O_CREATED', 'O_DECLINED', 'O_SELECTED', 'O_SENT', 'O_SENT_BACK', 'W_Afhandelen leads', 
'W_Beoordelen fraude', 'W_Completeren aanvraag', 'W_Nabellen incomplete dossiers', 'W_Nabellen offertes',
'W_Valideren aanvraag', 'W_Wijzigen contractgegevens']
features_activity = df_training[feature_names_activity]
features_activity_test = df_test[feature_names_activity]
def get_basic_model():
model = tf.keras.Sequential([
normalizer_activity,
tf.keras.layers.Dense(256, activation='relu'),
tf.keras.layers.Dense(256, activation='relu'),
tf.keras.layers.Dense(256, activation='relu'),
tf.keras.layers.Dense(256, activation='relu'),
tf.keras.layers.Dense(256, activation='relu'),
tf.keras.layers.Flatten(),
tf.keras.layers.Dense(24)
])

model.compile(optimizer='adam',
loss='sparse_categorical_crossentropy',
metrics=['accuracy'])
return model

model = get_basic_model()
model.fit(features_activity, target_activity, epochs=5, batch_size=181138)

我的目标是在DataFrame中添加一个预测列，为每一行(即df.loc[0][prediction] = predicted_activity(提供预测。如果我运行model.predict(df.loc[0][feature_names_activity])，我会得到一个24x33的表(其中24是不同活动的数量，33是功能的数量(。如何读取此表以获得预测，以及如何返回df的预测列？

Answer 1

您可以测试以下内容：

history = model.fit(features_activity, target_activity, epochs=5, batch_size=181138)
hist_df = pd.DataFrame(history.history) 
print(hist_df)

享受这个