def file_to_dicts(self, file: str) -> [dict]:
column_mapping = {
task["label_column_name"]: task["label_name"]
for task in self.tasks.values()
}
dicts = read_tsv(
rename_columns=column_mapping,
filename=file,
delimiter=self.delimiter,
skiprows=self.skiprows,
quotechar=self.quote_char,
)
# collect all labels and compute scaling stats
train_labels = []
for d in dicts:
train_labels.append(
float(d[self.tasks["regression"]["label_name"]]))
scaler = StandardScaler()
scaler.fit(np.reshape(train_labels, (-1, 1)))
# add to label list in regression task
self.tasks["regression"]["label_list"] = [
scaler.mean_.item(), scaler.scale_.item()
]
return dicts
def _file_to_dicts(self, file: str) -> [dict]:
dicts = read_tsv(
filename=file,
delimiter=self.delimiter,
skiprows=self.skiprows,
quotechar=self.quote_char,
)
return dicts
def _file_to_dicts(self, file: str) -> dict:
dicts = read_tsv(
filename=file,
delimiter=self.delimiter,
skiprows=self.skiprows,
columns=self.columns,
)
return dicts
def _file_to_dicts(self, file: str) -> [dict]:
column_mapping = {task["label_column_name"]: task["label_name"] for task in self.tasks.values()}
dicts = read_tsv(
filename=file,
delimiter=self.delimiter,
skiprows=self.skiprows,
quotechar=self.quote_char,
rename_columns=column_mapping,
header=self.header
)
return dicts
