I want to use classical machine learning models such XGBoost for my time series prediction. Since the input data for XGBoost/sklearn based models is 2d i.e. (n_samples, n_features), I want to encode positional/temporal information to the input data by following the approach of Transformer i.e. positional encoding. My question is how can we do this using the positional encoding concept of transformer? Does it make sense for time series prediction problem?