def main(args, uargs=None):
"""Run the ``catalyst-data split-dataframe`` script."""
dataframe = pd.read_csv(args.in_csv)
train_folds = (utils.folds_to_list(args.train_folds)
if args.train_folds is not None else None)
valid_folds = (utils.folds_to_list(args.valid_folds)
if args.valid_folds is not None else None)
infer_folds = (utils.folds_to_list(args.infer_folds)
if args.infer_folds is not None else None)
tag2class = (json.load(open(args.tag2class))
if args.tag2class is not None else None)
df_all, train, valid, infer = utils.split_dataframe(
dataframe,
train_folds=train_folds,
valid_folds=valid_folds,
infer_folds=infer_folds,
tag2class=tag2class,
tag_column=args.tag_column,
class_column=args.class_column,
seed=args.seed,
n_folds=args.num_folds,
)
out_csv: str = args.out_csv
if out_csv.endswith(".csv"):
out_csv = out_csv[:-4]
df_all.to_csv(f"{out_csv}.csv", index=False)
train.to_csv(f"{out_csv}_train.csv", index=False)
valid.to_csv(f"{out_csv}_valid.csv", index=False)
infer.to_csv(f"{out_csv}_infer.csv", index=False)
def main(args, uargs=None):
dataframe = pd.read_csv(args.in_csv)
train_folds = \
folds_to_list(args.train_folds) \
if args.train_folds is not None else None
valid_folds = \
folds_to_list(args.valid_folds) \
if args.valid_folds is not None else None
infer_folds = \
folds_to_list(args.infer_folds) \
if args.infer_folds is not None else None
tag2class = json.load(open(args.tag2class)) \
if args.tag2class is not None else None
df_all, train, valid, infer = split_dataframe(
dataframe,
train_folds=train_folds,
valid_folds=valid_folds,
infer_folds=infer_folds,
tag2class=tag2class,
tag_column=args.tag_column,
class_column=args.class_column,
seed=args.seed,
n_folds=args.num_folds)
out_csv: str = args.out_csv
if out_csv.endswith(".csv"):
out_csv = out_csv[:-4]
df_all.to_csv(f"{out_csv}.csv", index=False)
train.to_csv(f"{out_csv}_train.csv", index=False)
valid.to_csv(f"{out_csv}_valid.csv", index=False)
infer.to_csv(f"{out_csv}_infer.csv", index=False)