Time-series forecasting is ubiquitous in varied domains, comparable to retail, finance, manufacturing, healthcare and pure sciences. In retail use instances, for instance, it has been noticed that bettering demand forecasting accuracy can meaningfully cut back stock prices and enhance income. Deep studying (DL) fashions have emerged as a preferred strategy for forecasting wealthy, multivariate, time-series information as a result of they’ve confirmed to carry out effectively in quite a lot of settings (e.g., DL fashions dominated the M5 competitors leaderboard).
On the identical time, there was speedy progress in massive basis language fashions used for pure language processing (NLP) duties, comparable to translation, retrieval-augmented era, and code completion. These fashions are educated on large quantities of textual information derived from quite a lot of sources like frequent crawl and open-source code that enables them to establish patterns in languages. This makes them very highly effective zero-shot instruments; as an example, when paired with retrieval, they will reply questions on and summarize present occasions.
Regardless of DL-based forecasters largely outperforming conventional strategies and progress being made in decreasing coaching and inference prices, they face challenges: most DL architectures require lengthy and concerned coaching and validation cycles earlier than a buyer can take a look at the mannequin on a brand new time-series. A basis mannequin for time-series forecasting, in distinction, can present first rate out-of-the-box forecasts on unseen time-series information with no further coaching, enabling customers to deal with refining forecasts for the precise downstream process like retail demand planning.
To that finish, in “A decoder-only basis mannequin for time-series forecasting”, we introduce TimesFM, a single forecasting mannequin pre-trained on a big time-series corpus of 100 billion actual world time-points. In comparison with the newest massive language fashions (LLMs), TimesFM is way smaller (200M parameters), but we present that even at such scales, its zero-shot efficiency on quite a lot of unseen datasets of various domains and temporal granularities come near the state-of-the-art supervised approaches educated explicitly on these datasets. Later this yr we plan to make this mannequin accessible for exterior clients in Google Cloud Vertex AI.
A decoder-only basis mannequin for time-series forecasting
LLMs are normally educated in a decoder-only trend that entails three steps. First, textual content is damaged down into subwords referred to as tokens. Then, the tokens are fed into stacked causal transformer layers that produce an output corresponding to every enter token (it can not attend to future tokens). Lastly, the output comparable to the i-th token summarizes all the knowledge from earlier tokens and predicts the (i+1)-th token. Throughout inference, the LLM generates the output one token at a time. For instance, when prompted with “What’s the capital of France?”, it would generate the token “The”, then situation on “What’s the capital of France? The” to generate the following token “capital” and so forth till it generates the whole reply: “The capital of France is Paris”.
A basis mannequin for time-series forecasting ought to adapt to variable context (what we observe) and horizon (what we question the mannequin to forecast) lengths, whereas having sufficient capability to encode all patterns from a big pretraining dataset. Much like LLMs, we use stacked transformer layers (self-attention and feedforward layers) as the primary constructing blocks for the TimesFM mannequin. Within the context of time-series forecasting, we deal with a patch (a bunch of contiguous time-points) as a token that was popularized by a current long-horizon forecasting work. The duty then is to forecast the (i+1)-th patch of time-points given the i-th output on the finish of the stacked transformer layers.
Nevertheless, there are a number of key variations from language fashions. Firstly, we’d like a multilayer perceptron block with residual connections to transform a patch of time-series right into a token that may be enter to the transformer layers together with positional encodings (PE). For that, we use a residual block just like our prior work in long-horizon forecasting. Secondly, on the different finish, an output token from the stacked transformer can be utilized to foretell an extended size of subsequent time-points than the enter patch size, i.e., the output patch size could be bigger than the enter patch size.
Think about a time-series of size 512 time-points getting used to coach a TimesFM mannequin with enter patch size 32 and output patch size 128. Throughout coaching, the mannequin is concurrently educated to make use of the primary 32 time-points to forecast the following 128 time-points, the primary 64 time-points to forecast time-points 65 to 192, the primary 96 time-points to forecast time-points 97 to 224 and so forth. Throughout inference, suppose the mannequin is given a brand new time-series of size 256 and tasked with forecasting the following 256 time-points into the long run. The mannequin will first generate the long run predictions for time-points 257 to 384, then situation on the preliminary 256 size enter plus the generated output to generate time-points 385 to 512. However, if in our mannequin the output patch size was equal to the enter patch size of 32 then for a similar process we must undergo eight era steps as an alternative of simply the 2 above. This will increase the probabilities of extra errors accumulating and due to this fact, in follow, we see {that a} longer output patch size yields higher efficiency for long-horizon forecasting
 |
| TimesFM structure. |
Pretraining information
Identical to LLMs get higher with extra tokens, TimesFM requires a big quantity of respectable time collection information to study and enhance. We have now spent a terrific period of time creating and assessing our coaching datasets, and the next is what we’ve got discovered works finest:
Artificial information helps with the fundamentals. Significant artificial time-series information could be generated utilizing statistical fashions or bodily simulations. These fundamental temporal patterns can educate the mannequin the grammar of time collection forecasting.
Actual-world information provides real-world taste. We comb by accessible public time collection datasets, and selectively put collectively a big corpus of 100 billion time-points. Amongst these datasets there are Google Tendencies and Wikipedia Pageviews, which monitor what persons are all in favour of, and that properly mirrors tendencies and patterns in lots of different real-world time collection. This helps TimesFM perceive the larger image and generalize higher when supplied with domain-specific contexts not seen throughout coaching.
Zero-shot analysis outcomes
We consider TimesFM zero-shot on information not seen throughout coaching utilizing fashionable time-series benchmarks. We observe that TimesFM performs higher than most statistical strategies like ARIMA, ETS and might match or outperform highly effective DL fashions like DeepAR, PatchTST which were explicitly educated on the goal time-series.
We used the Monash Forecasting Archive to guage TimesFM’s out-of-the-box efficiency. This archive incorporates tens of hundreds of time-series from varied domains like site visitors, climate, and demand forecasting protecting frequencies starting from jiffy to yearly information. Following current literature, we examine the imply absolute error (MAE) appropriately scaled in order that it may be averaged throughout the datasets. We see that zero-shot (ZS) TimesFM is best than most supervised approaches, together with current deep studying fashions. We additionally examine TimesFM to GPT-3.5 for forecasting utilizing a selected prompting approach proposed by llmtime(ZS). We exhibit that TimesFM performs higher than llmtime(ZS) regardless of being orders of magnitude smaller.
 |
| Scaled MAE (the decrease the higher) of TimesFM(ZS) in opposition to different supervised and zero-shot approaches on Monash datasets. |
A lot of the Monash datasets are quick or medium horizon, i.e., the prediction size is just not too lengthy. We additionally take a look at TimesFM on fashionable benchmarks for lengthy horizon forecasting in opposition to a current state-of-the-art baseline PatchTST (and different long-horizon forecasting baselines). Within the subsequent determine, we plot the MAE on ETT datasets for the duty of predicting 96 and 192 time-points into the long run. The metric has been calculated on the final take a look at window of every dataset (as carried out by the llmtime paper). We see that TimesFM not solely surpasses the efficiency of llmtime(ZS) but additionally matches that of the supervised PatchTST mannequin explicitly educated on the respective datasets.
 |
| Final window MAE (the decrease the higher) of TimesFM(ZS) in opposition to llmtime(ZS) and long-horizon forecasting baselines on ETT datasets. |
Conclusion
We practice a decoder-only basis mannequin for time-series forecasting utilizing a big pretraining corpus of 100B actual world time-points, the vast majority of which was search curiosity time-series information derived from Google Tendencies and pageviews from Wikipedia. We present that even a comparatively small 200M parameter pretrained mannequin that makes use of our TimesFM structure shows spectacular zero-shot efficiency on quite a lot of public benchmarks from completely different domains and granularities.
Acknowledgements
This work is the results of a collaboration between a number of people throughout Google Analysis and Google Cloud, together with (in alphabetical order): Abhimanyu Das, Weihao Kong, Andrew Leach, Mike Lawrence, Alex Martin, Rajat Sen, Yang Yang and Yichen Zhou.