main_cols = ["id", "target", "comment_text"]
sub_target_cols = [
    "severe_toxicity", "obscene",
    "threat","insult",
    "identity_attack", "sexual_explicit",
]
identity_cols = [
    "male", "female",
    "homosexual_gay_or_lesbian",
    "christian", "jewish",
    "muslim", "black",
    "white", "psychiatric_or_mental_illness",
]

metadata_cols = [
    "created_date", "publication_id",
    "parent_id", "article_id",
    "rating", "funny",
    "wow", "sad",
    "likes", "disagree",
    "identity_annotator_count",
    "toxicity_annotator_count",
]
unused_identity = [
    col
    for col in df_train.columns
    if col not in main_cols + sub_target_cols + identity_cols + metadata_cols
]