Datasets:

nlpUc3mStudents

/

mental-risk-a

like 0

Cargar y concatenar los mesajes por conversación

ds = load_dataset('nlpUc3mStudents/mental-risk-a')
# to pandas
train_df = ds['train'].to_pandas()
# concat messages by subject id
train_by_subjectid = (
    train_df.groupby('subject_id')
    .agg({'message': lambda x: ' | '.join(x), 'label': 'first'})
    .reset_index()
    .assign(
        num_messages=lambda x: x.message.str.count('\|') + 1
    )
)
# back to datasets
ds_new = Dataset.from_pandas(train_by_subjectid)