module 'tensorflow_estimator.python.estimator.api._v1.estimator' has no attribute 'rnn'

Question

AttributeError: module 'tensorflow_estimator.python.estimator.api._v1.estimator' has no attribute 'rnn' keeps popping up. I am doing a project on Spelling correction and i really can't figure out why this error occured

Part of code where the error is shown:

def encoding_layer(rnn_size, sequence_length, num_layers, rnn_inputs, keep_prob, direction): '''Create the encoding layer'''

if direction == 1:
    with tf.name_scope("RNN_Encoder_Cell_1D"):
        for layer in range(num_layers):
            with tf.compat.v1.variable_scope('encoder_{}'.format(layer)):
                lstm = tf.compat.v1.estimator.rnn.LSTMCell(rnn_size)

                drop = tf.compat.v1.estimator.rnn.DropoutWrapper(lstm, 
                                                     input_keep_prob = keep_prob)
                enc_output, enc_state = tf.nn.dynamic_rnn(drop, 
                                                          rnn_inputs,
                                                          sequence_length,
                                                          dtype=tf.float32)

        return enc_output, enc_state
    
    
if direction == 2:
    with tf.name_scope("RNN_Encoder_Cell_2D"):
        for layer in range(num_layers):
            with tf.compat.v1.variable_scope('encoder_{}'.format(layer)):
                cell_fw = tf.compat.v1.estimator.rnn.LSTMCell(rnn_size)
                cell_fw = tf.compat.v1.estimator.rnn.DropoutWrapper(cell_fw, 
                                                        input_keep_prob = keep_prob)

                cell_bw = tf.contrib.rnn.LSTMCell(rnn_size)
                cell_bw = tf.contrib.rnn.DropoutWrapper(cell_bw, 
                                                        input_keep_prob = keep_prob)

                enc_output, enc_state = tf.nn.bidirectional_dynamic_rnn(cell_fw, 
                                                                        cell_bw, 
                                                                        rnn_inputs,
                                                                        sequence_length,
                                                                        dtype=tf.float32)
        # Join outputs since we are using a bidirectional RNN
        enc_output = tf.concat(enc_output,2)
        # Use only the forward state because the model can't use both states at once
       return enc_output, enc_state[0]