---
title: Blending gradient boosted trees and neural networks for point and probabilistic forecasting of hierarchical time series
url: https://www.ml-quant.com/papers/doi/10-1016-j-ijforecast-2022-01-001/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-09-26
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: doi:10.1016/j.ijforecast.2022.01.001
source_url: http://dx.doi.org/10.1016/j.ijforecast.2022.01.001
featured: 2023-10-25
citations: 17
topic: Econometrics & Forecasting
---


# Blending gradient boosted trees and neural networks for point and probabilistic forecasting of hierarchical time series

The paper outlines a successful method for point and probabilistic forecasting using a mix of machine learning models, as demonstrated in the M5 Competition, highlighting the significance of diverse models and careful validation example selection.

- Source: http://dx.doi.org/10.1016/j.ijforecast.2022.01.001
- Identifier: doi:10.1016/j.ijforecast.2022.01.001
- Released: 2023-10-19
- First featured: Quant Letter No. 23 (2023-10-25): https://www.ml-quant.com/issues/2023-10-25/
- Citations (Semantic Scholar): 17
- Published in: International Journal of Forecasting
- Topic: Econometrics & Forecasting

## Related

- [TKAN: Temporal Kolmogorov-Arnold Networks](https://www.ml-quant.com/papers/ssrn/4825654/): The article presents Temporal Kolomogorov-Arnold Networks (TKANs), a new neural network design that merges the benefits of Recurrent Neural Networks and Long Short-Term Memory for improved multistep time series forecasting.
- [Augmented HAR](https://www.ml-quant.com/papers/ssrn/4516177/): The Augmented HAR algorithm, combined with artificial neural networks, enhances forecast accuracy for stocks with less than seven years of data.
- [Universality of Linear Recurrences Followed by Non-linear Projections: Finite-Width Guarantees and Benefits of Complex Eigenvalues](https://www.ml-quant.com/papers/arxiv/2307.11888/): The study reveals that deep neural networks using linear complex-valued RNNs and MLPs can accurately approximate regular causal sequence-to-sequence maps, with complex eigenvalues near unit disk aiding in information storage.
- [Diffusive Gibbs Sampling](https://www.ml-quant.com/papers/arxiv/2402.03008/): The article introduces Diffusive Gibbs Sampling (DiGS), a new method for sampling from multi-modal distributions, which performs better in tasks like Bayesian neural networks and molecular dynamics.
- [Decomposing weather forecasting into advection and convection with neural networks](https://www.ml-quant.com/papers/arxiv/2405.06590/): The research introduces a machine learning model for weather forecasting that separately learns horizontal and vertical atmospheric movements, surpassing existing methods in accuracy and efficiency.
- [Distributional Refinement Network: Distributional Forecasting via Deep Learning](https://www.ml-quant.com/papers/arxiv/2406.00998/): The article introduces the Distributional Refinement Network (DRN), a model that enhances predictive performance and interpretability in actuarial modelling by merging a baseline model with a flexible neural network.