cleanup TODO list

Author: carusyte

corl/wc_data/input_fn.py

@@ -631,6 +631,7 @@ def getWorkloadForPredictionFromTags(corl_prior, max_step, time_shift, host, por
     '''
     Returns list of tuples (code, date, klid)
     '''
+    # TODO need to compare corl_prior and max_step with snapshot values in db (secu.params), and opt to re-initialize tag table if 2 versions differ.
     global cnxpool
     _init_db(1, host, port, pwd)
     cnx = cnxpool.get_connection()

corl/wcc/worker.py

@@ -40,7 +40,6 @@
 
 
 def _init(db_pool_size=None, db_host=None, db_port=None, db_pwd=None):
-    # FIXME too many db initialization message in the log and 'aborted clients' in mysql dashboard
     global cnxpool
     size = db_pool_size or 5
     print("{} initializing mysql connection pool of size {}".format(

pstk/data/data15.py

@@ -1,5 +1,3 @@
-
-
 import os
 import sys
 import psutil
@@ -10,6 +8,8 @@
 from time import strftime
 from mysql.connector.pooling import MySQLConnectionPool
 
+#TODO establish data pipeline prototype for stock trend prediction
+
 k_cols = ["lr"]
 
 idxlst = None

pstk/model/model2.py

@@ -39,7 +39,7 @@ def wrapper(self):
 
 
 def length(data):
-    # with tf.variable_scope("rnn_length"):    #FIXME no scope?
+    # with tf.variable_scope("rnn_length"):    
     used = tf.sign(tf.reduce_max(input_tensor=tf.abs(data), axis=2))
     length = tf.reduce_sum(input_tensor=used, axis=1)
     length = tf.cast(length, tf.int32)
@@ -52,7 +52,7 @@ def fusedBN(input, scale, offset, mean, variance, training):
 
 
 def last_relevant(output, length):
-    # with tf.variable_scope("rnn_last"):       #FIXME no scope?
+    # with tf.variable_scope("rnn_last"):       
     batch_size = tf.shape(input=output)[0]
     max_length = int(output.get_shape()[1])
     output_size = int(output.get_shape()[2])
@@ -64,19 +64,19 @@ def last_relevant(output, length):
 
 def conv2d(input, filters, seq):
     conv = tf.compat.v1.layers.conv2d(
-        # name="conv_lv{}".format(seq),    #FIXME perhaps no name?
+        # name="conv_lv{}".format(seq),    
         inputs=input,
         filters=filters,
         kernel_size=2,
         kernel_initializer=tf.compat.v1.truncated_normal_initializer(
             stddev=0.01),
         bias_initializer=tf.compat.v1.constant_initializer(0.1),
         padding="same",
-        activation=tf.nn.elu)  # FIXME or perhaps relu6??
+        activation=tf.nn.elu)  
     h_stride = 2 if int(conv.get_shape()[1]) >= 2 else 1
     w_stride = 2 if int(conv.get_shape()[2]) >= 2 else 1
     pool = tf.compat.v1.layers.max_pooling2d(
-        # name="pool_lv{}".format(seq),    #FIXME perhaps no name?
+        # name="pool_lv{}".format(seq),    
         inputs=conv, pool_size=2, strides=[h_stride, w_stride],
         padding="same")
     # can't use tf.nn.batch_normalization in a mapped function
@@ -119,33 +119,33 @@ def prediction(self):
         rnn = self.rnn(self, cnn)
         dense = tf.compat.v1.layers.dense(
             inputs=rnn,
-            units=self._num_hidden * 3,  # FIXME fallback to 3 * hidden size?
+            units=self._num_hidden * 3,  
             kernel_initializer=tf.compat.v1.truncated_normal_initializer(stddev=0.01),
             bias_initializer=tf.compat.v1.constant_initializer(0.1),
-            activation=tf.nn.elu)  # FIXME sure elu?
+            activation=tf.nn.elu)  
         dropout = tf.compat.v1.layers.dropout(
             inputs=dense, rate=0.5, training=self.training)
         output = tf.compat.v1.layers.dense(
             inputs=dropout,
             units=int(self.target.get_shape()[1]),
             kernel_initializer=tf.compat.v1.truncated_normal_initializer(stddev=0.01),
             bias_initializer=tf.compat.v1.constant_initializer(0.1),
-            activation=tf.nn.relu6)  # FIXME fall back to relu6?
+            activation=tf.nn.relu6)  
         return output
 
     @staticmethod
     def rnn(self, input):
         # Recurrent network.
         cells = []
-        state_size = self._num_hidden  # FIXME fallback to 128
+        state_size = self._num_hidden  
         for _ in range(self._num_layers):
             # Or LSTMCell(num_units), or use ConvLSTMCell?
             cell = tf.compat.v1.nn.rnn_cell.GRUCell(
                 state_size,
                 kernel_initializer=tf.compat.v1.truncated_normal_initializer(
                     stddev=0.01),
                 bias_initializer=tf.compat.v1.constant_initializer(0.1))
-            # activation=None)  # FIXME fall back to None?
+            # activation=None)  
             cells.append(cell)
         cell = tf.compat.v1.nn.rnn_cell.MultiRNNCell(cells)
         _length = length(input)
@@ -167,7 +167,7 @@ def cnn2d(input, training, height):
         accepts input shape: [step_size, time_shift*features]
         transformed to: [step_size, time_shift(height), features(width), channel]
         """
-        # with tf.variable_scope("conv2d_parent"):    #FIXME no scope?
+        # with tf.variable_scope("conv2d_parent"):    
         print("shape of cnn input: {}".format(input.get_shape()))
         width = int(input.get_shape()[1])//height
         input2d = tf.reshape(input, [-1, height, width, 1])
@@ -191,17 +191,17 @@ def cnn2d(input, training, height):
         convlayer = tf.squeeze(convlayer, [1, 2])
         print("squeeze: {}".format(convlayer.get_shape()))
         dense = tf.compat.v1.layers.dense(
-            # name="cnn2d_dense",     #FIXME no name?
+            # name="cnn2d_dense",     
             inputs=convlayer,
             units=convlayer.get_shape()[1]*2,
             kernel_initializer=tf.compat.v1.truncated_normal_initializer(
                 stddev=0.01),
             bias_initializer=tf.compat.v1.constant_initializer(0.1),
-            activation=tf.nn.elu  # FIXME or perhaps elu?
+            activation=tf.nn.elu  
         )
         print("dense: {}".format(dense.get_shape()))
         dropout = tf.compat.v1.layers.dropout(
-            # name="cnn2d_dropout",    #FIXME no name?
+            # name="cnn2d_dropout",    
             inputs=dense, rate=0.5, training=training)
         return dropout
 

test/test24_mdnc.py

@@ -3,10 +3,8 @@
 from model import dnc_regressor
 from common.common import next_power_of_2
 # Path hack.
-import sys
-import os
 import numpy as np
-sys.path.append(os.path.dirname(os.path.abspath(__file__)) + "/..")
+# sys.path.append(os.path.dirname(os.path.abspath(__file__)) + "/..")
 
 # pylint: disable-msg=E0401
 
@@ -78,6 +76,7 @@ def create_regressor():
 
 
 if __name__ == '__main__':
+    # TODO add data pipeline, make it fully runnable
 
     np.random.seed(SEED)