Some Recent Advances in Forecasting and Control

By G. Box, G. Jenkins, John F. MacGregor. 1968

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Tian Zhou
2. Abhimanyu Das
3. Zhijian Xu
4. Ailing Zeng
5. Haixu Wu
6. Shiyu Wang
7. Mingsheng Long
8. Jianmin Wang

Citations in the corpus

• DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks (2017, methodology): “ossibly admitting probabilistic forecasts through simulation or closed-form expressions for the predictive distributions. Many methods in this class are based on the classical Box-Jenkins methodology [3], exponential smoothing techniques, or state space models [11, 19]. In recent years, a new type o”
• Think Globally, Act Locally: A Deep Neural Network Approach to High-Dimensional Time Series Forecasting (2019, methodology): “For a comprehensive treatment of traditional methods, we refer the readers to [13, 19, 4, 18, 12] and the references there in.”
• ARMDN: Associative and Recurrent Mixture Density Networks for eRetail Demand Forecasting (2018, methodology): “Traditional time series methods like Box-Jenkins [4, 5] and variations of the state space models like ARIMA and Holt-Winters[14], have been used for demand forecasting in the industry for a very long time.”
• Multi-Horizon Time Series Forecasting with Temporal Attention Learning (2019, unknown): “ClassicaltimeseriesforecastingapproachesincludeHolt-Winters method [15, 30] and ARIMA [5].”
• SCINet: Time Series Modeling and Forecasting with Sample Convolution and Interaction (2021, methodology): “Traditional time series forecasting methods such as the autoregressive integrated moving average (ARIMA) model [8] and Holt-Winters seasonal method [14] have theoretical guarantees.”
• FiLM: Frequency improved Legendre Memory Model for Long-term Time Series Forecasting (2022, background): “Since classic models such as ARIMA and simple RNN/TCN models perform inferior as shown in Zhou et al. (2021) and Wu et al. (2021), we mainly include five state-of-the-art (SOTA) Transformer-based models, i.e., FEDformer, Autoformer Wu et al. (2021), Informer Zhou et al. (2021), Log-Trans Li et al. (2”
• One Fits All: Power General Time Series Analysis by Pretrained LM (2023, unknown): “The first generation of well-discussed models can be dated back to auto-regressive family, such as ARIMA Box & Jenkins (1968); Box & Pierce (1970)”
• iTransformer: Inverted Transformers Are Effective for Time Series Forecasting (2023, background): “…the numerical but less semantic relationship among time points, researchers find that simple linear layers, which can be traced back to statistical forecasters (Box & Jenkins, 1968), have exceeded complicated Transformers on both performance and efficiency (Zeng et al., 2023; Das et al., 2023).”
• Long-term Forecasting with TiDE: Time-series Dense Encoder (2023, methodology): “1984) and the Box-Jenkins methodology (Box and Jenkins, 1968) belong in this category.”
• Position-Based Content Attention for Time Series Forecasting with Sequence-to-Sequence RNNs (2017, background): “Notable among these are autoregressive (AR) [9] and moving averages (MA) [10] models, that were combined in a more general and effective framework, known as autoregressive moving average (ARMA), or autoregressive integrated moving average (ARIMA) when the differencing is included in the model [11].”
• A decoder-only foundation model for time-series forecasting (2023, methodology): “In recent times, Deep learning models (Salinas et al., 2020; Ore-shkin et al., 2019) have emerged as a popular approach for forecasting rich, multivariate, time series data, often out-performing classical statistical approaches such as ARIMA or GARCH (Box & Jenkins, 1968).”
• Non-stationary Transformers: Exploring the Stationarity in Time Series Forecasting (2022, methodology): “For univariate forecasting, we include seven competitive baselines: N-HiTS [9], N-BEATS [27], Autoformer [37], Pyraformer [23], Informer [39], Reformer [19] and ARIMA [7].”
• Analysis of photovoltaic system performance time series: Seasonality and performance loss (2015, methodology): “More specifically, seasonal adjustment techniques originating from econometrics, such as classical seasonal decomposition (CSD) [14,15] and regression modeling with Auto-Regressive Integrated Moving Average (ARIMA) [16,17] errors (regARIMA) [18], are used to decompose a seasonal time series into the”
• Large Language Models Are Zero-Shot Time Series Forecasters (2023, methodology): “For Darts, we use several meth-ods that are implemented directly in the package, including neural network models (TCN [28], N-BEATS [37], N-HiTS [11]) and simple moving average models (ARIMA [8]).”
• Pyraformer: Low-Complexity Pyramidal Attention for Long-Range Time Series Modeling and Forecasting (2022, background): “The first group involves ARIMA (Box & Jenkins, 1968) and Prophet (Taylor & Letham, 2018).”
• FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecasting (2022, methodology): “To the contrary of classic models like Autoregressive integrated moving average (ARIMA) which has a predefined data bias structure for output distribution, Transformer-based models forecast each point independently and solely based on the overall MSE loss learning.”