def test_fit(self):
domain = DateTimeDomain()
X = numpy.array([
"1960-01-01T00:00:00", "1960-01-01T00:00:01",
"1960-01-01T00:01:00", "1959-12-31T23:59:59", "1960-01-03T03:30:03"
])
Xt = domain.fit_transform(X)
self.assertEqual([
datetime(1960, 1, 1, 0, 0, 0),
datetime(1960, 1, 1, 0, 0, 1),
datetime(1960, 1, 1, 0, 1, 0),
datetime(1959, 12, 31, 23, 59, 59),
datetime(1960, 1, 3, 3, 30, 3)
], Xt.tolist())
def test_timedelta_seconds(self):
X = DataFrame([["2018-12-31T23:59:59", "2019-01-01T00:00:00"], ["2019-01-01T03:30:03", "2019-01-01T00:00:00"]], columns = ["left", "right"])
mapper = DataFrameMapper([
(["left", "right"], [DateTimeDomain(), SecondsSinceYearTransformer(year = 2010), ExpressionTransformer("X[0] - X[1]")])
])
Xt = mapper.fit_transform(X)
self.assertEqual([[-1], [12603]], Xt.tolist())
def test_timedelta(self): X = DataFrame([["2020-01-01"], ["2020-01-01T00:01:00"], ["2020-01-01T05:23:30"]], columns = ["timestamp"]) mapper = DataFrameMapper([ (["timestamp"], [DateTimeDomain(), SecondsSinceMidnightTransformer()]) ]) Xt = mapper.fit_transform(X) self.assertEqual([[0], [60], [19410]], Xt.tolist())
def test_fit_transform(self):
X = DataFrame([["1959-12-31", "1959-12-31T23:59:59"],
["1960-01-01", "1960-01-01T01:01:10"],
["2003-04-01", "2003-04-01T05:16:27"]],
columns=["date", "datetime"])
domain = clone(DateDomain())
Xt = domain.fit_transform(X)
self.assertEqual([
datetime(1959, 12, 31),
datetime(1960, 1, 1),
datetime(2003, 4, 1)
], Xt["date"].tolist())
self.assertEqual([
datetime(1959, 12, 31),
datetime(1960, 1, 1),
datetime(2003, 4, 1)
], Xt["datetime"].tolist())
Xt = domain.fit_transform(X.values)
self.assertEqual([
datetime(1959, 12, 31),
datetime(1960, 1, 1),
datetime(2003, 4, 1)
], Xt[:, 0].tolist())
domain = clone(DateTimeDomain())
Xt = domain.fit_transform(X)
self.assertEqual([
datetime(1959, 12, 31, 0, 0, 0),
datetime(1960, 1, 1, 0, 0, 0),
datetime(2003, 4, 1, 0, 0, 0)
], Xt["date"].tolist())
self.assertEqual([
datetime(1959, 12, 31, 23, 59, 59),
datetime(1960, 1, 1, 1, 1, 10),
datetime(2003, 4, 1, 5, 16, 27)
], Xt["datetime"].tolist())
Xt = domain.fit_transform(X.values)
self.assertEqual([
datetime(1959, 12, 31, 0, 0, 0),
datetime(1960, 1, 1, 0, 0, 0),
datetime(2003, 4, 1, 0, 0, 0)
], Xt[:, 0].tolist())
store_csv(df, "Apollo")
def build_apollo(mapper, name):
pipeline = PMMLPipeline([("mapper", mapper),
("classifier", DecisionTreeClassifier())])
pipeline.fit(df, df["success"])
store_pkl(pipeline, name)
success = DataFrame(pipeline.predict(df), columns=["success"])
success_proba = DataFrame(
pipeline.predict_proba(df),
columns=["probability(false)", "probability(true)"])
success = pandas.concat((success, success_proba), axis=1)
store_csv(success, name)
mapper = DataFrameMapper([(["launch", "return"], [
DateTimeDomain(),
DaysSinceYearTransformer(year=1968),
ExpressionTransformer("X[1] - X[0]")
])])
build_apollo(mapper, "DurationInDaysApollo")
mapper = DataFrameMapper([(["launch", "return"], [
DateTimeDomain(),
SecondsSinceYearTransformer(year=1968),
ExpressionTransformer("X[1] - X[0]")
])])
build_apollo(mapper, "DurationInSecondsApollo")