ML之LiR&DNN&EL：基于skflow的LiR、DNN、sklearn的RF对Boston(波士顿房价)数据集进行回归预测(房价)

输出结果

设计思路

核心代码

tf_lr = skflow.TensorFlowLinearRegressor(steps=10000, learning_rate=0.01, batch_size=50)

tf_lr.fit(X_train, y_train)  

tf_lr_y_predict = tf_lr.predict(X_test)

tf_dnn_regressor = skflow.TensorFlowDNNRegressor(hidden_units=[100, 40],

   steps=10000, learning_rate=0.01, batch_size=50)

tf_dnn_regressor.fit(X_train, y_train)

tf_dnn_regressor_y_predict = tf_dnn_regressor.predict(X_test)

rfr = RandomForestRegressor()

rfr.fit(X_train, y_train)

rfr_y_predict = rfr.predict(X_test)

class TensorFlowLinearRegressor(TensorFlowEstimator, RegressorMixin):

   """TensorFlow Linear Regression model."""

   def __init__(self, n_classes=0, tf_master="", batch_size=32, steps=200, optimizer="SGD",

       learning_rate=0.1, tf_random_seed=42, continue_training=False,

       num_cores=4, verbose=1, early_stopping_rounds=None,

       max_to_keep=5, keep_checkpoint_every_n_hours=10000):

       super(TensorFlowLinearRegressor, self).__init__(model_fn=models.linear_regression,

        n_classes=n_classes, tf_master=tf_master, batch_size=batch_size, steps=steps,

        optimizer=optimizer, learning_rate=learning_rate, tf_random_seed=tf_random_seed,

        continue_training=continue_training, num_cores=num_cores, verbose=verbose,

        early_stopping_rounds=early_stopping_rounds, max_to_keep=max_to_keep,

        keep_checkpoint_every_n_hours=keep_checkpoint_every_n_hours)

   @property

   def weights_(self):

       """Returns weights of the linear regression."""

       return self.get_tensor_value('linear_regression/weights:0')

   @property

   def bias_(self):

       """Returns bias of the linear regression."""

       return self.get_tensor_value('linear_regression/bias:0')