def test_learning_tracker_scheduled_should_output_dates_after_feedback():
lt = LearningTracker(epoch=datetime.now())
lt.learned(result=0.40, when=0.01)
lt.learned(result=0.44, when=0.8)
for offset in lt.schedule_as_offset(stop=10):
assert offset > 0.8
def test_learning_tracker_broken_feedback_curve_reproduction():
lt = LearningTracker(epoch=datetime.now())
lt.learned(result=0.44, when=0.8)
lt.learned(result=0.64, when=1.75)
lt.learned(result=0.55, when=1.3)
lt.learned(result=0.40, when=0.001)
lt.learned(result=0.64, when=3.02)
lt.plot_graphs(stop=30)
lt.save_figure("results/lt_broken_feedback_curve.pdf")
def test_serialization_epoch():
lt = LearningTracker(epoch=datetime.now(), )
moments, results = get_feedback1()
for index, (moment, result) in enumerate(zip(moments[0:4], results)):
lt.learned(when=moment, result=result)
byte_stream = pickle.dumps(lt)
unpickled_learning_tracker = pickle.loads(byte_stream)
hdl, _ = unpickled_learning_tracker.plot_graphs(stop=40)
unpickled_learning_tracker.save_figure("results/post_pickle.pdf")
hdl.close()
def test_learning_tracker_longterm_response():
start_time = datetime.now()
moments, results = get_feedback1()
lt = LearningTracker(epoch=start_time)
for index, (moment, result) in enumerate(zip(moments[0:2], results)):
lt.learned(when=moment, result=result)
hd, _ = lt.plot_graphs(stop=300)
lt.save_figure("results/space_learning_tracker_too_long.pdf")
hd.close()
def test_learning_tracker_broken_two_handle_graph():
lt = LearningTracker(epoch=datetime.now())
# give our learning tracker some feedback
for d, r in zip(
[0, 0.8, 1.75, 3.02, 4.8, 7.33],
[0.40, 0.44, 0.64, 0.76, 0.83, 0.89],
):
lt.learned(result=r, when=d)
with lt.graphs(stop=43,
show=True,
control_handle=True,
reference_handle=True,
filename='results/context_manager_two_handles.svg') as (rh,
ch):
r_m = lt.reference.range_for(curve=3, stop=43, day_step_size=0.5)
r_p = [lt.reference.predict_result(moment, curve=3) for moment in r_m]
rh.plot(r_m, r_p, color='xkcd:ruby')
c_m = lt.range_for(curve=1, stop=43, day_step_size=0.5)
c_p = [lt.predict_result(moment, curve=1) for moment in c_m]
ch.plot(c_m, c_p, color='xkcd:azure')
def test_learning_tracker_next_offset_feature():
lt = LearningTracker(epoch=datetime.now())
lt.learned(result=0.40, when=0.001)
lt.learned(result=0.44, when=0.8)
lt.learned(result=0.55, when=1.3)
lt.learned(result=0.64, when=1.75)
lt.learned(result=0.76, when=3.02)
results = [
[4.8, 7.33, 10.93, 16.00, 23.00, 29.00],
[0.83, 0.89, 1.00, 0.99, 0.99, 1.00],
]
for d, r in zip(*results):
days_since_training_epoch, result = d, r
lt.learned(result=result, when=days_since_training_epoch)
sc = lt.next_offset()
assert abs(sc - 122.4043) < 0.01