def test_apply():
s1 = Stream([1, 4, 9, 16, 25], "s1")
lag = s1.apply(np.sqrt).rename("apply")
feed = DataFeed([lag])
feed.compile()
values = []
while feed.has_next():
values += [feed.next()["apply"]]
assert values == [1, 2, 3, 4, 5]
def test_fillna():
s = Stream([1, np.nan, -3, np.nan, 5], "s")
m = s.fillna(-1).rename("fill")
feed = DataFeed([m])
feed.compile()
expected = [1, -1, -3, -1, 5]
actual = []
while feed.has_next():
actual += [feed.next()["fill"]]
assert actual == expected
def test_clamp_max():
s = Stream([1, -2, -3, 0, 5], "s")
m = s.clamp_max(0).rename("clamp_max")
feed = DataFeed([m])
feed.compile()
expected = [0, -2, -3, 0, 0]
actual = []
while feed.has_next():
actual += [feed.next()["clamp_max"]]
assert actual == expected
def test_lag():
s1 = Stream([1, 2, 3, 4, 5]).rename("stream")
assert s1.name == "stream"
lag = s1.lag()
assert lag.name == "Lag(stream,1)"
feed = DataFeed([lag])
feed.compile()
values = []
while feed.has_next():
values += [feed.next()["Lag(stream,1)"]]
assert values == [np.nan, 1, 2, 3, 4]
def test_max():
s1 = Stream([1, 2, 3, 4], "s1")
s2 = Stream([1, 4, 3, 2], "s2")
m = s1.max(s2).rename("max")
feed = DataFeed([m])
feed.compile()
expected = [1, 4, 3, 4]
actual = []
while feed.has_next():
actual += [feed.next()["max"]]
assert actual == expected
def test_emwmv_unbiased():
# bias: True
v = [np.nan, 2, np.nan, 6, 8, 5]
s = Stream(v, "s")
specs = [{
"alpha": 0.68,
"adjust": True,
"ignore_na": True,
"min_periods": 3
}, {
"alpha": 0.68,
"adjust": True,
"ignore_na": False,
"min_periods": 3
}, {
"alpha": 0.68,
"adjust": False,
"ignore_na": True,
"min_periods": 3
}, {
"alpha": 0.68,
"adjust": False,
"ignore_na": False,
"min_periods": 3
}]
for spec in specs:
d = spec.copy()
d["warmup"] = d["min_periods"]
del d["min_periods"]
var = s.ewm(**d).var(bias=False).rename("var")
feed = DataFeed([var])
feed.compile()
expected = list(pd.Series(v).ewm(**spec).var(bias=False))
actual = []
while feed.has_next():
actual += [feed.next()["var"]]
assert all(np.isclose(actual, expected))
def test_sub():
s1 = Stream([2, 3, 4, 5, 6], "s1")
assert s1.name == "s1"
s2 = Stream([1, 2, 3, 4, 5], "s2")
assert s2.name == "s2"
s = s1 - s2
assert s.name == "Subtract(s1,s2)"
feed = DataFeed([s])
feed.compile()
values = []
while feed.has_next():
values += [feed.next()["Subtract(s1,s2)"]]
assert values == [1, 1, 1, 1, 1]
def test_add():
s1 = Stream([1, 2, 3, 4, 5], "s1")
assert s1.name == "s1"
s2 = Stream([1, 2, 3, 4, 5], "s2")
assert s2.name == "s2"
s = s1 + s2
assert s.name == "Add(s1,s2)"
feed = DataFeed([s])
feed.compile()
values = []
while feed.has_next():
values += [feed.next()["Add(s1,s2)"]]
assert values == [2, 4, 6, 8, 10]
def test_ewa_beginning_na():
# adjust: True, ignore_na: False
v = [np.nan, 2, np.nan, 6, 8, 5]
s = Stream(v).rename("s")
specs = [{
"alpha": 0.68,
"adjust": True,
"ignore_na": True,
"min_periods": 3
}, {
"alpha": 0.68,
"adjust": True,
"ignore_na": False,
"min_periods": 3
}, {
"alpha": 0.68,
"adjust": False,
"ignore_na": True,
"min_periods": 3
}, {
"alpha": 0.68,
"adjust": False,
"ignore_na": False,
"min_periods": 3
}]
for spec in specs:
d = spec.copy()
d["warmup"] = d["min_periods"]
del d["min_periods"]
mean = s.ewm(**d).mean().rename("mean")
feed = DataFeed([mean])
feed.compile()
expected = list(pd.Series(v).ewm(**spec).mean())
actual = []
while feed.has_next():
actual += [feed.next()["mean"]]
assert all(np.isclose(actual, expected))
def test_log_returns():
s1 = Stream([200.23, 198.35, 244.36, 266.30, 250.40], "price")
assert s1.name == "price"
lp = s1.log()
lr = lp - lp.lag()
feed = DataFeed([lr])
feed.compile()
while feed.has_next():
print(feed.next())
lr = s1.log().diff().rename("log_return")
feed = DataFeed([lr])
feed.compile()
while feed.has_next():
print(feed.next())
def test_truediv():
s1 = Stream([2, 3, 4, 5, 6], "s1")
s2 = Stream([2, 3, 4, 5, 6], "s2")
s = (s1 / s2).rename("div")
feed = DataFeed([s])
feed.compile()
values = []
while feed.has_next():
values += [feed.next()["div"]]
assert values == [1, 1, 1, 1, 1]
s1 = Stream([1, 1, 1, 1, 1], "s2")
s = (5 / s1).rename("div")
feed = DataFeed([s])
feed.compile()
values = []
while feed.has_next():
values += [feed.next()["div"]]
assert values == [5, 5, 5, 5, 5]
s1 = Stream([5, 5, 5, 5, 5], "s2")
s = (s1 / 5).rename("div")
feed = DataFeed([s])
feed.compile()
values = []
while feed.has_next():
values += [feed.next()["div"]]
assert values == [1, 1, 1, 1, 1]
def test_ewma():
# adjust: True, ignore_na: True
v = [5, 2, 4, 6]
s = Stream(v, "s")
mean = s.ewm(alpha=0.68, adjust=True, ignore_na=True).mean().rename("mean")
feed = DataFeed([mean])
feed.compile()
expected = list(
pd.Series(v).ewm(alpha=0.68, adjust=True, ignore_na=True).mean())
actual = []
while feed.has_next():
actual += [feed.next()["mean"]]
assert all(np.isclose(actual, expected))
# adjust: True, ignore_na: False
v = [5, 2, np.nan, 6]
s = Stream(v, "s")
mean = s.ewm(alpha=0.68, adjust=True,
ignore_na=False).mean().rename("mean")
feed = DataFeed([mean])
feed.compile()
expected = list(
pd.Series(v).ewm(alpha=0.68, adjust=True, ignore_na=False).mean())
actual = []
while feed.has_next():
actual += [feed.next()["mean"]]
assert all(np.isclose(actual, expected))
# adjust: True, ignore_na: False
v = [5, 2, np.nan, 6]
s = Stream(v, "s")
mean = s.ewm(alpha=0.68, adjust=False,
ignore_na=True).mean().rename("mean")
feed = DataFeed([mean])
feed.compile()
expected = list(
pd.Series(v).ewm(alpha=0.68, adjust=False, ignore_na=True).mean())
actual = []
while feed.has_next():
actual += [feed.next()["mean"]]
assert all(np.isclose(actual, expected))
# adjust: True, ignore_na: False
v = [5, 2, np.nan, 6]
s = Stream(v, "s")
mean = s.ewm(alpha=0.68, adjust=False,
ignore_na=False).mean().rename("mean")
feed = DataFeed([mean])
feed.compile()
expected = list(
pd.Series(v).ewm(alpha=0.68, adjust=False, ignore_na=False).mean())
actual = []
while feed.has_next():
actual += [feed.next()["mean"]]
assert all(np.isclose(actual, expected))