Abstract: State space models for time series forecasting have been dominated by the single source of error (SSOE) framework, most notably the ETS family of models. The idea of SSOE is to use the same error across all equations in the model. Multiple source of error (MSOE) models, by contrast, assign independent stochastic disturbances to each component – level, trend, and seasonality – offering a richer and more flexible representation of uncertainty, yet they lack a systematic, unifying taxonomy. This paper introduces the PTS taxonomy, a structured classification of MSOE state space models tailored to the specific properties of the MSOE setting. The taxonomy organises models along three dimensions: P (Power transform, based on the Box-Cox transformation), T (Trend, with options for none, local, global, or damped), and S (Seasonality, with options for none, discrete, or trigonometric), yielding up to 24 well-defined model variants. All models are cast within a general state space system and estimated via the Kalman filter using maximum likelihood, with model selection performed through standard information criteria. The framework also incorporates a robust outlier detection procedure covering additive outliers, level shifts, and slope changes, as well as natural handling of missing observations through the Kalman smoother. We illustrate the practical utility of the taxonomy through empirical experiments on the real life dataset, demonstrating that the PTS family is both theoretically coherent and empirically competitive with established alternatives.

Slides are here.
Package that implements PTS: muse

Message ISF2026: PTS Taxonomy of Multiple Source of Error State Space Models for Demand Forecasting first appeared on Open Forecast.

As for my contribution, I presented a paper that Nikos Kourentzes and I have been working on since around 2018. It focuses on pooling using point information criteria. The core idea is to combine forecasts based on a smaller pool of models, which we propose creating by comparing the distributions of information criteria across forecasting models. We’re planning to finish a new version of the paper by September and submit it to a peer-reviewed journal. I’ll share more details when the draft that I can share is ready. In the meantime, you can check out the slides that summarise the main points of the paper. Here they are.

Message ITISE2025: Beyond summary performance metrics for forecast selection and combination first appeared on Open Forecast.

First things first, the event was organised by Mitchell O’Hara, Xiaoqian Wang, Bahman Rostami-Tabar, Azul Garza, Resul Akay, Shanika Wickramasuriya and me. My contribution was acting as the programme chair of the workshop and helping invite some speakers. The event was sponsored by the International Institute of Forecasters, and you can find some information about it on the following website. The workshop lasted for two days, and there were several great talks delivered by excellent speakers (see the full programme here).

But among those, there were two talks especially important to me personally, because they were related to the smooth package:

1. by Kandrika Pritularga, explaining the core of the package, the maths behind it, and how it works,
2. by Filotas Theodosiou, who explained his excellent work on translating the smooth R code to Python.

Yes, you’ve heard correctly! We have finally had some progress translating the smooth package from R to Python (done between Filotas, Leonidas Tsaprounis and me. I’m a vibe coder now! :D). The Python code is available in this GitHub branch, and we now even have a description of the package structure and function flow (thanks to Fil and his AI friends).

It’s still a work in progress, and there’s lots to do, but it’s an important step for us. We plan to continue working on it, and there are already several people who have started helping us with the translation. So, stay tuned – great times are coming!

Message IIF Open Source Forecasting software workshop and smooth first appeared on Open Forecast.

My presentation was based on this paper. And here are the slides:
2025-05-01-IMA-OR

Message 5th IMA and OR Society Conference first appeared on Open Forecast.

Abstract: Bootstrap is extensively used in statistics and machine learning for cross-sectional data to account for uncertainty about the data, model form, and parameter estimates. However, conventional methods may not be suitable for time series data due to autocorrelation and specific dynamic structures. Over the years, various approaches have been developed to address this issue. Some assume specific models (e.g., STL), while others are non-parametric (e.g., Maximum Entropy Bootstrap, MEB). However, the former can be overly restrictive, while the latter may not perform well in case of outliers and external drivers. To address these issues, we propose a non-parametric bootstrap approach inspired by MEB, which does not assume any structure in the data yet creates reasonable copies of existing time series of different nature. These copies can be utilised in bagged ETS/ARIMA or any other approach involving small sample uncertainty. We demonstrate how the proposed bootstrap works using real-time series examples and assess improvements it brings in terms of forecasting accuracy compared to conventional approaches.

Here are the slides of the presentation.

And here is me, trying not to attract attention of statisticians:

How to attract attention of statisticians…

Message ISF2024: How to Bootstrap Time Series without Attracting Attention of Statisticians first appeared on Open Forecast.

ETS+ARIMA love story with happy ending…

This talk was based on Chapter 9 of ADAM monograph, and more specifically on Section 9.4.

The slides of the presentation are available here.

Message ISF2022: How to make ETS work with ARIMA first appeared on Open Forecast.

Presentation at ISF2021 Lancaster Hub

I presented on the topic of Multiplicative ETS, based on this chapter of the ADAM textbook and on the paper John Boylan and I are working on. In this presentation, I discuss that the point forecasts from the ETS(M,*,*) models in general do not correspond to conditional expectations (and not even to geometric means) and show the difference between the two. Furthermore, we propose using Log Normal, Gamma and Inverse Gaussian distributions for the error term of the model instead of the Normal one. This makes the model more realistic and help in forecasting on low volume data. All of this is already implemented in adam() function of smooth package for R.

Here are the slides.

Message ISF2021: How to Make Multiplicative ETS Work for You first appeared on Open Forecast.

The slides of the presentation are available here.

Message Multi-step Estimators and Shrinkage Effect in Time Series Models – presentation for CEBA first appeared on Open Forecast.

Retail sales of tomatoes

Anyway, here are the slides if anyone wants to have a look.

Message International Symposium on Forecasting 2019 first appeared on Open Forecast.

1. State space ARIMA for Supply Chain Forecasting, based on which I have developed a module for Smoothie a couple of years ago,
2. Artificial Intelligence in Business, one of the modern hot topics that the company wanted to know a little bit more about.

Presentation at SMUG2019

The conference was interesting, showing what the company does and what it stands for. They are doing a good job in developing the software for forecasting and inventory control and supporting their users. Plus, I finally had a chance to meet in person with both founders of the company (Bill Tonetti and Eric Townson), as well as with the other members of their team. Overall, it was a pleasant experience and an interesting event.

As for the presentations, they seemed to go well, and the participants of the conference looked satisfied. Here are the slides:

1. SMUG2019 – Svetunkov – ARIMA
2. SMUG2019 – Svetunkov – AI in Business

Message SMUG2019 first appeared on Open Forecast.