def define_inputs(self) -> tp.Dict[str, trac.ModelInputSchema]:
customer_loans = trac.declare_input_table(
trac.F("id",
trac.BasicType.STRING,
label="Customer account ID",
business_key=True),
trac.F("loan_amount",
trac.BasicType.DECIMAL,
label="Principal loan amount",
format_code="CCY:EUR"),
trac.F("total_pymnt",
trac.BasicType.DECIMAL,
label="Total amount repaid",
format_code="CCY:EUR"),
trac.F("region",
trac.BasicType.STRING,
label="Customer home region",
categorical=True),
trac.F("loan_condition_cat",
trac.BasicType.INTEGER,
label="Loan condition category",
categorical=True))
return {"customer_loans": customer_loans}
def define_outputs(self) -> tp.Dict[str, _api.ModelOutputSchema]:
profit_by_region = _api.declare_output_table(
_api.F("region", _api.BasicType.STRING, label="Customer home region", categorical=True),
_api.F("gross_profit", _api.BasicType.DECIMAL, label="Total gross profit", format_code="CCY:USD"))
return {"profit_by_region": profit_by_region}
def define_outputs(self) -> tp.Dict[str, trac.ModelOutputSchema]:
preprocessed = trac.declare_output_table(
trac.F("id", trac.BasicType.STRING, label="Customer account ID", business_key=True),
trac.F("some_quantity_x", trac.BasicType.DECIMAL, label="Some quantity X", format_code="CCY:EUR"))
return {"preprocessed_data": preprocessed}
def define_outputs(self) -> tp.Dict[str, trac.ModelOutputSchema]:
output1 = trac.declare_output_table(
trac.F("field1", trac.BasicType.INTEGER,
"Something about this field"),
trac.F("field2", trac.BasicType.FLOAT,
"Something about this other field"))
return {"output1": output1}
def define_outputs(self) -> tp.Dict[str, api.ModelOutputSchema]:
output_table_1 = api.declare_output_table(
api.F("output_field_1",
api.STRING,
label="Output field 1",
business_key=True),
api.F("output_field_2", api.DATE, label="Output field 2"),
api.F("output_field_3", api.FLOAT, label="Output field 3"))
return {"output_table_1": output_table_1}
def define_inputs(self) -> tp.Dict[str, api.ModelInputSchema]:
input_table_1 = api.declare_input_table(
api.F("input_field_1",
api.STRING,
label="Input field 1",
business_key=True),
api.F("input_field_2", api.INTEGER, label="Input field 2"),
api.F("input_field_3",
api.STRING,
label="Input field 3",
categorical=True),
api.F("input_field_4", api.DECIMAL, label="Input field 4"))
return {"input_table_1": input_table_1}
def define_inputs(self) -> tp.Dict[str, trac.ModelInputSchema]:
customer_loans = trac.declare_input_table(
trac.F("id", trac.BasicType.STRING, label="Customer account ID", business_key=True),
trac.F("loan_amount", trac.BasicType.DECIMAL, label="Principal loan amount", format_code="CCY:EUR"),
trac.F("total_pymnt", trac.BasicType.DECIMAL, label="Total amount repaid", format_code="CCY:EUR"),
trac.F("region", trac.BasicType.STRING, label="Customer home region", categorical=True),
trac.F("loan_condition_cat", trac.BasicType.INTEGER, label="Loan condition category", categorical=True))
currency_data = trac.declare_input_table(
trac.F("ccy_code", trac.BasicType.STRING, label="Currency code", categorical=True),
trac.F("spot_date", trac.BasicType.DATE, label="Spot date for FX rate"),
trac.F("dollar_rate", trac.BasicType.DECIMAL, label="Dollar FX rate", format_code="CCY:USD"))
return {"customer_loans": customer_loans, "currency_data": currency_data}
class DataRoundTripModel(trac.TracModel):
ROUND_TRIP_FIELDS = [
trac.F("boolean_field", trac.BasicType.BOOLEAN, label="BOOLEAN field"),
trac.F("integer_field", trac.BasicType.INTEGER, label="INTEGER field"),
trac.F("float_field", trac.BasicType.FLOAT, label="FLOAT field"),
trac.F("decimal_field", trac.BasicType.DECIMAL, label="DECIMAL field"),
trac.F("string_field", trac.BasicType.STRING, label="STRING field"),
trac.F("date_field", trac.BasicType.DATE, label="DATE field"),
trac.F("datetime_field", trac.BasicType.DATETIME, label="DATETIME field")]
def define_parameters(self) -> tp.Dict[str, trac.ModelParameter]:
return trac.declare_parameters(
trac.P("use_spark", trac.BasicType.BOOLEAN, default_value=False, label="Use Spark for round trip testing"))
def define_inputs(self) -> tp.Dict[str, trac.ModelInputSchema]:
round_trip_input = trac.declare_input_table(self.ROUND_TRIP_FIELDS)
return {"round_trip_input": round_trip_input}
def define_outputs(self) -> tp.Dict[str, trac.ModelOutputSchema]:
round_trip_output = trac.declare_input_table(self.ROUND_TRIP_FIELDS)
return {"round_trip_output": round_trip_output}
def run_model(self, ctx: trac.TracContext):
use_spark = ctx.get_parameter("use_spark")
if use_spark:
round_trip_input = ctx.get_spark_table("round_trip_input")
ctx.put_spark_table("round_trip_output", round_trip_input)
else:
round_trip_input = ctx.get_pandas_table("round_trip_input")
ctx.put_pandas_table("round_trip_output", round_trip_input)