Machine learning methods for GEFCom2017 probabilistic load forecasting

This paper describes the preprocessing and forecasting methods used by team Orbuculum during the qualifying match of the Global Energy Forecasting Competition 2017 (GEFCom2017). Tree-based algorithms (gradient boosting and quantile random forest) and neural networks made up an ensemble. The ensemble prediction quantiles were obtained by a simple averaging of the ensemble members’ prediction quantiles. The result shows a robust performance according to the pinball loss metric, with the ensemble model achieving third place in the qualifying match of the competition.     

An ensemble approach to GEFCom2017 probabilistic load forecasting

We present an ensembling approach to medium-term probabilistic load forecasting which ranked second out of 73 competitors in the defined data track of the GEFCom2017 qualifying match. In addition to being accurate, the ensemble method is highly scalable, due to the fact that it had to be applied to nine quantiles in ten zones and for six rounds. Candidate forecasts were generated using random settings for input data, covariates, and learning algorithms. The best candidate forecasts were averaged to create the final forecast, with the number of candidate forecasts being chosen based on their accuracy in similar validation periods.     

Neural networks for GEFCom2017 probabilistic load forecasting

This report describes the forecasting model which was developed by team “4C” for the global energy forecasting competition 2017 (GEFCom2017), with some modifications added afterwards to improve its accuracy. The model is based on neural networks. Temperature scenarios obtained from historical data are used as inputs to the neural networks in order to create load scenarios, and these load scenarios are then transformed into quantiles. By using a feature selection approach that is based on a stepwise regression technique, a neural network based model is developed for each zone. Furthermore, a dynamic choice of the temperature scenarios is suggested. The feature selection and dynamic choice of the temperature scenarios can improve the quantile scores considerably, resulting in very accurate forecasts among the top teams.     

Data visualization and forecast combination for probabilistic load forecasting in GEFCom2017 final match

This paper describes the methods used by Team Cassandra, a joint effort between IBM Research Australia and the University of Melbourne, in the GEFCom2017 load forecasting competition. An important first phase in the forecasting effort involved a deep exploration of the underlying dataset. Several data visualisation techniques were applied to help us better understand the nature and size of gaps, outliers, the relationships between different entities in the dataset, and the relevance of custom date ranges. Improved, cleaned data were then used to train multiple probabilistic forecasting models. These included a number of standard and well-known approaches, as well as a neural-network based quantile forecast model that was developed specifically for this dataset. Finally, model selection and forecast combination were used to choose a custom forecasting model for every entity in the dataset.     

Kaggle forecasting competitions: An overlooked learning opportunity

We review the results of six forecasting competitions based on the online data science platform Kaggle, which have been largely overlooked by the forecasting community. In contrast to the M competitions, the competitions reviewed in this study feature daily and weekly time series with exogenous variables, business hierarchy information, or both. Furthermore, the Kaggle data sets all exhibit higher entropy than the M3 and M4 competitions, and they are intermittent.In this review, we confirm the conclusion of the M4 competition that ensemble models using cross-learning tend to outperform local time series models and that gradient boosted decision trees and neural networks are strong forecast methods. Moreover, we present insights regarding the use of external information and validation strategies, and discuss the impacts of data characteristics on the choice of statistics or machine learning methods. Based on these insights, we construct nine ex-ante hypotheses for the outcome of the M5 competition to allow empirical validation of our findings.     

Quantile regression for the qualifying match of GEFCom2017 probabilistic load forecasting

We present a simple quantile regression-based forecasting method that was applied in the probabilistic load forecasting framework of the Global Energy Forecasting Competition 2017 (GEFCom2017). The hourly load data are log transformed and split into a long-term trend component and a remainder term. The key forecasting element is the quantile regression approach for the remainder term, which takes into account both weekly and annual seasonalities, such as their interactions. Temperature information is used only for stabilizing the forecast of the long-term trend component. Information on public holidays is ignored. However, the forecasting method still placed second in the open data track and fourth in the definite data track, which is remarkable given the simplicity of the model. The method also outperforms the Vanilla benchmark consistently.     

Reconciled boosted models for GEFCom2017 hierarchical probabilistic load forecasting

When forecasting time series in a hierarchical configuration, it is necessary to ensure that the forecasts reconcile at all levels. The 2017 Global Energy Forecasting Competition (GEFCom2017) focused on addressing this topic. Quantile forecasts for eight zones and two aggregated zones in New England were required for every hour of a future month. This paper presents a new methodology for forecasting quantiles in a hierarchy which outperforms a commonly-used benchmark model. A simulation-based approach was used to generate demand forecasts. Adjustments were made to each of the demand simulations to ensure that all zonal forecasts reconciled appropriately, and a weighted reconciliation approach was implemented to ensure that the bottom-level zonal forecasts summed correctly to the aggregated zonal forecasts. We show that reconciling in this manner improves the forecast accuracy. A discussion of the results and modelling performances is presented, and brief reviews of hierarchical time series forecasting and gradient boosting are also included.     

Rethinking weather station selection for electric load forecasting using genetic algorithms

Demand forecasting is and has been for years a topic of great interest in the electricity sector, being the temperature one of its major drivers. Indeed, one of the challenges when modelling the load is to choose the right weather station, or set of stations, for a given load time series. However, only a few research papers have been devoted to this topic. This paper reviews the most relevant methods that were applied during the Global Energy Forecasting Competition of 2014 (GEFCom2014) and presents a new approach to weather station selection, based on Genetic Algorithms (GA), which allows finding the best set of stations for any demand forecasting model, and outperforms the results of existing methods. Furthermore its performance has also been tested using GEFCom2012 data, providing significant error improvements. Finally, the possibility of combining the weather stations selected by the proposed GA using the BFGS algorithm is briefly tested, providing promising results.     

A refined parametric model for short term load forecasting

We present a refined parametric model for forecasting electricity demand which performed particularly well in the recent Global Energy Forecasting Competition (GEFCom 2012). We begin by motivating and presenting a simple parametric model, treating the electricity demand as a function of the temperature and day of the data. We then set out a series of refinements of the model, explaining the rationale for each, and using the competition scores to demonstrate that each successive refinement step increases the accuracy of the model’s predictions. These refinements include combining models from multiple weather stations, removing outliers from the historical data, and special treatments of public holidays.     

Global Energy Forecasting Competition 2012

The Global Energy Forecasting Competition (GEFCom2012) attracted hundreds of participants worldwide, who contributed many novel ideas to the energy forecasting field. This paper introduces both tracks of GEFCom2012, hierarchical load forecasting and wind power forecasting, with details on the aspects of the problem, the data, and a summary of the methods used by selected top entries. We also discuss the lessons learned from this competition from the organizers’ perspective. The complete data set, including the solution data, is published along with this paper, in an effort to establish a benchmark data pool for the community.     

The M5 competition: Background,organization, and implementation

The M5 competition follows the previous four M competitions, whose purpose is to learn from empirical evidence how to improve forecasting performance and advance the theory and practice of forecasting. M5 focused on a retail sales forecasting application with the objective to produce the most accurate point forecasts for 42,840 time series that represent the hierarchical unit sales of the largest retail company in the world, Walmart, as well as to provide the most accurate estimates of the uncertainty of these forecasts. Hence, the competition consisted of two parallel challenges, namely the Accuracy and Uncertainty forecasting competitions. M5 extended the results of the previous M competitions by: (a) significantly expanding the number of participating methods, especially those in the category of machine learning; (b) evaluating the performance of the uncertainty distribution along with point forecast accuracy; (c) including exogenous/explanatory variables in addition to the time series data; (d) using grouped, correlated time series; and (e) focusing on series that display intermittency. This paper describes the background, organization, and implementations of the competition, and it presents the data used and their characteristics. Consequently, it serves as introductory material to the results of the two forecasting challenges to facilitate their understanding.     

The tourism forecasting competition

We evaluate the performances of various methods for forecasting tourism data. The data used include 366 monthly series, 427 quarterly series and 518 annual series, all supplied to us by either tourism bodies or academics who had used them in previous tourism forecasting studies. The forecasting methods implemented in the competition are univariate and multivariate time series approaches, and econometric models. This forecasting competition differs from previous competitions in several ways: (i) we concentrate on tourism data only; (ii) we include approaches with explanatory variables; (iii) we evaluate the forecast interval coverage as well as the point forecast accuracy; (iv) we observe the effect of temporal aggregation on the forecasting accuracy; and (v) we consider the mean absolute scaled error as an alternative forecasting accuracy measure. We find that pure time series approaches provide more accurate forecasts for tourism data than models with explanatory variables. For seasonal data we implement three fully automated pure time series algorithms that generate accurate point forecasts, and two of these also produce forecast coverage probabilities which are satisfactorily close to the nominal rates. For annual data we find that Naïve forecasts are hard to beat.     

Correlated daily time series and forecasting in the M4 competition

We participated in the M4 competition for time series forecasting and here describe our methods for forecasting daily time series. We used an ensemble of five statistical forecasting methods and a method that we refer to as the correlator. Our retrospective analysis using the ground truth values published by the M4 organisers after the competition demonstrates that the correlator was responsible for most of our gains over the naïve constant forecasting method. We identify data leakage as one reason for its success, due partly to test data selected from different time intervals, and partly to quality issues with the original time series. We suggest that future forecasting competitions should provide actual dates for the time series so that some of these leakages could be avoided by participants.     

Forecast combinations of computational intelligence and linear models for the NN5 time series forecasting competition

In this work we introduce the forecasting model with which we participated in the NN5 forecasting competition (the forecasting of 111 time series representing daily cash withdrawal amounts at ATM machines). The main idea of this model is to utilize the concept of forecast combination, which has proven to be an effective methodology in the forecasting literature. In the proposed system we attempted to follow a principled approach, and make use of some of the guidelines and concepts that are known in the forecasting literature to lead to superior performance. For example, we considered various previous comparison studies and time series competitions as guidance in determining which individual forecasting models to test (for possible inclusion in the forecast combination system). The final model ended up consisting of neural networks, Gaussian process regression, and linear models, combined by simple average. We also paid extra attention to the seasonality aspect, decomposing the seasonality into weekly (which is the strongest one), day of the month, and month of the year seasonality.     

Hierarchical forecasting with a top-down alignment of independent-level forecasts

Hierarchical forecasting with intermittent time series is a challenge in both research and empirical studies. Extensive research focuses on improving the accuracy of each hierarchy, especially the intermittent time series at bottom levels. Then, hierarchical reconciliation can be used to improve the overall performance further. In this paper, we present a hierarchical-forecasting-with-alignment approach that treats the bottom-level forecasts as mutable to ensure higher forecasting accuracy on the upper levels of the hierarchy. We employ a pure deep learning forecasting approach, N-BEATS, for continuous time series at the top levels, and a widely used tree-based algorithm, LightGBM, for intermittent time series at the bottom level. The hierarchical-forecasting-with-alignment approach is a simple yet effective variant of the bottom-up method, accounting for biases that are difficult to observe at the bottom level. It allows suboptimal forecasts at the lower level to retain a higher overall performance. The approach in this empirical study was developed by the first author during the M5 Accuracy competition, ranking second place. The method is also business orientated and can be used to facilitate strategic business planning.     

Data preprocessing and quantile regression for probabilistic load forecasting in the GEFCom2017 final match

Team QUINKAN competed in the GEFCom2017 final match of hierarchical probabilistic load forecasting by adopting the quantile regression method using the R package quantreg. The weather stations were clustered into 11 groups, from which an optimal one was chosen for each load meter using the boosting method. The load meter records were cleaned and/or supplemented by various methods in order to secure robust quantile predictions. The variation in the regression formulas was kept as small as possible by introducing measures for suppressing prediction instability, although special formulas were employed for loading meters that were of an industrial nature. Several procedures were applied to help improve the accuracy, such as the smoothing of season transitions, coarse graining of the relative humidity, the use of load-oriented day-type definition, the averaging of weather data, and outlier removal.     

The performance of the global bottom-up approach in the M5 accuracy competition: A robustness check

The M5 accuracy competition has presented a large-scale hierarchical forecasting problem in a realistic grocery retail setting in order to evaluate an extended range of forecasting methods, particularly those adopting machine learning. The top ranking solutions adopted a global bottom-up approach, by which is meant using global forecasting methods to generate bottom level forecasts in the hierarchy and then using a bottom-up strategy to obtain coherent forecasts for aggregate levels. However, whether the observed superior performance of the global bottom-up approach is robust over various test periods or only an accidental result, is an important question for retail forecasting researchers and practitioners. We conduct experiments to explore the robustness of the global bottom-up approach, and make comments on the efforts made by the top-ranking teams to improve the core approach. We find that the top-ranking global bottom-up approaches lack robustness across time periods in the M5 data. This inconsistent performance makes the M5 final rankings somewhat of a lottery. In future forecasting competitions, we suggest the use of multiple rolling test sets to evaluate the forecasting performance in order to reward robustly performing forecasting methods, a much needed characteristic in any application.     

Out-of-sample tests of forecasting accuracy: an analysis and review

In evaluations of forecasting accuracy, including forecasting competitions, researchers have paid attention to the selection of time series and to the appropriateness of forecast-error measures. However, they have not formally analyzed choices in the implementation of out-of-sample tests, making it difficult to replicate and compare forecasting accuracy studies. In this paper, I (1) explain the structure of out-of-sample tests, (2) provide guidelines for implementing these tests, and (3) evaluate the adequacy of out-of-sample tests in forecasting software. The issues examined include series-splitting rules, fixed versus rolling origins, updating versus recalibration of model coefficients, fixed versus rolling windows, single versus multiple test periods, diversification through multiple time series, and design characteristics of forecasting competitions. For individual time series, the efficiency and reliability of out-of-sample tests can be improved by employing rolling-origin evaluations, recalibrating coefficients, and using multiple test periods. The results of forecasting competitions would be more generalizable if based upon precisely described groups of time series, in which the series are homogeneous within group and heterogeneous between groups. Few forecasting software programs adequately implement out-of-sample evaluations, especially general statistical packages and spreadsheet add-ins.     

Parameter-efficient deep probabilistic forecasting

Probabilistic time series forecasting is crucial in many application domains, such as retail, ecommerce, finance, and biology. With the increasing availability of large volumes of data, a number of neural architectures have been proposed for this problem. In particular, Transformer-based methods achieve state-of-the-art performance on real-world benchmarks. However, these methods require a large number of parameters to be learned, which imposes high memory requirements on the computational resources for training such models. To address this problem, we introduce a novel bidirectional temporal convolutional network that requires an order of magnitude fewer parameters than a common Transformer-based approach. Our model combines two temporal convolutional networks: the first network encodes future covariates of the time series, whereas the second network encodes past observations and covariates. We jointly estimate the parameters of an output distribution via these two networks. Experiments on four real-world datasets show that our method performs on par with four state-of-the-art probabilistic forecasting methods, including a Transformer-based approach and WaveNet, on two point metrics (sMAPE and NRMSE) as well as on a set of range metrics (quantile loss percentiles) in the majority of cases. We also demonstrate that our method requires significantly fewer parameters than Transformer-based methods, which means that the model can be trained faster with significantly lower memory requirements, which as a consequence reduces the infrastructure cost for deploying these models.     

Online hierarchical forecasting for power consumption data

This paper proposes a three-step approach to forecasting time series of electricity consumption at different levels of household aggregation. These series are linked by hierarchical constraints—global consumption is the sum of regional consumption, for example. First, benchmark forecasts are generated for all series using generalized additive models. Second, for each series, the aggregation algorithm ML-Poly, introduced by Gaillard, Stoltz, and van Erven in 2014, finds an optimal linear combination of the benchmarks. Finally, the forecasts are projected onto a coherent subspace to ensure that the final forecasts satisfy the hierarchical constraints. By minimizing a regret criterion, we show that the aggregation and projection steps improve the root mean square error of the forecasts. Our approach is tested on household electricity consumption data; experimental results suggest that successive aggregation and projection steps improve the benchmark forecasts at different levels of household aggregation.