class TaskPreprocess(d6tflow.tasks.TaskPqPandas):
do_preprocess = d6tflow.BoolParameter(default=True) # parameter for preprocessing yes/no
def run(self):
df_train = self.input().load() # quickly load required data
if self.do_preprocess:
df_train.iloc[:,:-1] = sklearn.preprocessing.scale(df_train.iloc[:,:-1])
self.save(df_train)
class ModelFeatures(d6tflow.tasks.TaskPqPandas):
interaction = d6tflow.BoolParameter(default=False)
def run(self):
df = self.inputLoad()
df['target'] = df['sales']
if self.interaction:
df['tv_radio'] = df['TV'] * df['radio']
self.save(df)
class TaskPreprocess(d6tflow.tasks.TaskPqPandas):
do_preprocess = d6tflow.BoolParameter(
default=cfg.do_preprocess) # parameter for preprocessing yes/no
def run(self):
df_train = self.input().load(
) # quickly load required data, see https://d6tflow.readthedocs.io/en/latest/tasks.html#load-input-data
if self.do_preprocess:
df_train.iloc[:, :-1] = sklearn.preprocessing.scale(
df_train.iloc[:, :-1])
self.save(
df_train
) # save task output, see https://d6tflow.readthedocs.io/en/latest/tasks.html#save-output-data