def main(self, ds):
df = ds.get("titanic").content
df[self.column] = df[self.column].fillna(df[self.column].median())
rds = DataSet()
rds.put(f"fillna_{self.column}", DataFrameData(df[self.column]))
return rds
def main(self, ds):
df = ds.get("titanic").content
df.loc[df["Sex"] == "male", "Sex"] = 0
df.loc[df["Sex"] == "female", "Sex"] = 1
rds = DataSet()
rds.put("sex_to_code", DataFrameData(df["Sex"]))
return rds
def main(self, ds):
df = ds.get("titanic").content
df.loc[df["Sex"] == "male", "Sex"] = 0
df.loc[df["Sex"] == "female", "Sex"] = 1
ds = DataSet()
ds.put("titanic", DataFrameData(df))
return ds
def main(self, ds):
df = ds.get("titanic").content
df["Embarked"] = df["Embarked"].fillna("S")
df.loc[df["Embarked"] == "S", "Embarked"] = 0
df.loc[df["Embarked"] == "C", "Embarked"] = 1
df.loc[df["Embarked"] == "Q", "Embarked"] = 2
rds = DataSet()
rds.put("embarked_to_code", DataFrameData(df["Embarked"]))
return rds
def main(self, ds):
df = ds.get("titanic").content
df["Sex"][df["Sex"] == "male"] = 0
df["Sex"][df["Sex"] == "female"] = 1
rds = DataSet()
rds.put("sex_to_code", DataFrameData(df["Sex"]))
time.sleep(random.randint(3, 10))
return rds
def main(self, ds):
df = ds.get("titanic").content
df = df.drop(["Age", "Sex", "Embarked"], axis=1)
df = df.join(ds.get("fillna_Age").content)
df = df.join(ds.get("sex_to_code").content)
df = df.join(ds.get("embarked_to_code").content)
rds = DataSet()
rds.put("titanic_result", DataFrameData(df))
return rds
def main(self, ds):
df = ds.get("titanic").content
df["Embarked"] = df["Embarked"].fillna("S")
df["Embarked"][df["Embarked"] == "S"] = 0
df["Embarked"][df["Embarked"] == "C"] = 1
df["Embarked"][df["Embarked"] == "Q"] = 2
rds = DataSet()
rds.put("embarked_to_code", DataFrameData(df["Embarked"]))
time.sleep(random.randint(3, 10))
return rds
def test_dataframe(self):
repo = LocalFileRepository(Path(os.path.dirname(__file__)) / Path("titanic.csv"))
data = DataFrameData.load(repo)
db_repo = PandasDbRepository(self.engine, "titanic")
dfd = DataFrameData(data.content, db_repo)
dfd.save()
dfd2 = DataFrameData.load(db_repo)
self.assertIsNotNone(dfd2.content)
self.assertTrue(data.content.equals(dfd2.content))