src/graph/temporal_split.py

import numpy as np
import pandas as pd


def temporal_split(df: pd.DataFrame, train_ratio=0.7, val_ratio=0.15):
    df = df.sort_values("timestamp").reset_index(drop=True)

    # time thresholds (CRITICAL)
    t_train = df["timestamp"].quantile(train_ratio)
    t_val = df["timestamp"].quantile(train_ratio + val_ratio)

    train_mask = df["timestamp"] <= t_train
    val_mask = (df["timestamp"] > t_train) & (df["timestamp"] <= t_val)
    test_mask = df["timestamp"] > t_val

    return train_mask, val_mask, test_mask, t_train