import datetime
import tensorflow as tf
class TensorBoard(object):
"""
TensorBoard is a visualization tool provided with TensorFlow.
This class can be used to record attributes from a running
Zipline algorithm.
If you have installed TensorFlow with pip, you should be able
to launch TensorBoard from the command line:
tensorboard --logdir=/full_path_to_your_logs
Args:
log_dir: the path of the directory where to save the log
files to be parsed by tensorboard
max_queue: Maximum number of summaries or events pending to be
written to disk before one of the 'add' calls block.
[default 10]
flush_secs: How often, in seconds, to flush the added summaries
and events to disk. [default 120]
"""
def __init__(self, log_dir='./logs', max_queue=10, flush_secs=120):
self.log_dir = log_dir
self.merged = tf.merge_all_summaries()
self.writer = tf.train.SummaryWriter(self.log_dir,
max_queue=max_queue,
flush_secs=flush_secs,
graph_def=None)
def log_dict(self, epoch, logs):
"""
Writes a dictionary of simple named values to TensorBoard.
Args:
epoch: An integer representing time.
logs: A dict containing what we want to log to TensorBoard.
"""
for name, value in logs.items():
summary = tf.Summary()
summary_value = summary.value.add()
summary_value.simple_value = value
summary_value.tag = name
self.writer.add_summary(summary, global_step=epoch)
self.writer.flush()
def log_algo(self, algo, epoch=None, other_logs={}):
"""
Logs info about a Zipline algorithm as it's running.
Args:
epoch: An integer representing algorithm time.
If None, then the algorithm's current
date is converted to an ordinal so that
these integers are monotonically increasing
with time. The same integer convention should
be used across different runs so that their charts
line up correctly.
algo: An instance of a zipline.algorithm.TradingAlgorithm
other_logs: A dictionary containing other things we want to log.
"""
if epoch is None:
epoch = datetime.date.toordinal(algo.get_datetime())
logs = {}
# add portfolio related things
logs['portfolio value'] = algo.portfolio.portfolio_value
logs['portfolio pnl'] = algo.portfolio.pnl
logs['portfolio return'] = algo.portfolio.returns
logs['portfolio cash'] = algo.portfolio.cash
logs['portfolio capital used'] = algo.portfolio.capital_used
logs['portfolio positions exposure'] = algo.portfolio.positions_exposure
logs['portfolio positions value'] = algo.portfolio.positions_value
logs['number of orders'] = len(algo.blotter.orders)
logs['number of open orders'] = len(algo.blotter.open_orders)
logs['number of open positions'] = len(algo.portfolio.positions)
# add recorded variables from `zipline.algorithm.record` method
for name, value in algo.recorded_vars.items():
logs[name] = value
# add any extras passed in through `other_logs` dictionary
for name, value in other_logs.items():
logs[name] = value
self.log_dict(epoch, logs)
