def test_transform_dom_fmt_list(self):
df = pd.DataFrame(
{"date_column": ["2016-05-28", "2016-06-27", "2016-07-26"]})
trainable = Map(columns=[day_of_month(it.date_column, "%Y-%m-%d")])
trained = trainable.fit(df)
transformed_df = trained.transform(df)
self.assertEqual(transformed_df["date_column"][0], 28)
self.assertEqual(transformed_df["date_column"][1], 27)
self.assertEqual(transformed_df["date_column"][2], 26)
def test_transform_dom_fmt_list(self):
df = pd.DataFrame(
{"date_column": ["28/05/2016", "27/06/2016", "26/07/2016"]})
sdf = self.sqlCtx.createDataFrame(df)
trainable = Map(columns=[day_of_month(it.date_column, "d/M/y")])
trained = trainable.fit(sdf)
transformed_df = trained.transform(sdf)
self.assertEqual(transformed_df.collect()[0]["date_column"], 28)
self.assertEqual(transformed_df.collect()[1]["date_column"], 27)
self.assertEqual(transformed_df.collect()[2]["date_column"], 26)
def test_transform_dom_list(self):
df = pd.DataFrame(
{"date_column": ["2016-05-28", "2016-06-27", "2016-07-26"]})
sdf = self.sqlCtx.createDataFrame(df)
trainable = Map(columns=[day_of_month(it.date_column)])
trained = trainable.fit(sdf)
transformed_df = trained.transform(sdf)
self.assertEqual(transformed_df.collect()[0]["date_column"], 28)
self.assertEqual(transformed_df.collect()[1]["date_column"], 27)
self.assertEqual(transformed_df.collect()[2]["date_column"], 26)
def test_transform_dom_fmt_map(self):
df = pd.DataFrame(
{"date_column": ["2016-05-28", "2016-06-27", "2016-07-26"]})
sdf = self.sqlCtx.createDataFrame(df)
trainable = Map(columns={"dom": day_of_month(it.date_column, "y-M-d")})
trained = trainable.fit(sdf)
transformed_df = trained.transform(sdf)
self.assertEqual((transformed_df.count(), len(transformed_df.columns)),
(3, 1))
self.assertEqual(transformed_df.collect()[0]["dom"], 28)
self.assertEqual(transformed_df.collect()[1]["dom"], 27)
self.assertEqual(transformed_df.collect()[2]["dom"], 26)