Time series forecasting
A method for time series forecasting includes receiving a time series forecasting query from a user requesting the data processing hardware to perform a plurality of time series forecasts. Each time series forecast is a forecast of future data based on respective current data Simultaneously, for each time series forecast of the plurality of time series forecasts requested by the time series forecasting query, the method includes training a plurality of models for the respective time series forecast. The method also includes determining which model of the plurality of models best fits the respective time series forecast and forecasting the future data based on the determined best fitting model and the respective current data. The method also includes returning, to the user, the forecasted future data for each of the plurality of time series forecasts request by the timer series forecasting query.
1. A computer-implemented method when executed by data processing hardware causes the data processing hardware to perform operations comprising:
receiving a time series forecasting query from a user device, the time series forecasting query requesting the data processing hardware to perform a plurality of time series forecasts, each time series forecast being a forecast of future data based on respective current data;
for each time series forecast of the plurality of time series forecasts requested by the time series forecasting query, simultaneously:
training a plurality of models for the respective time series forecast decomposed into multiple components;
for each respective model of the plurality of models, estimating, by the data processing hardware, using the respective current data, a respective relative quality of the respective model, the respective relative quality representing a quality of the respective model relative to each other model of the plurality of models;
selecting, using the respective relative quality of each respective model without relying on empirical means, one of the plurality of models that best fits the respective time series forecast; and
forecasting the future data based on the selected best fitting model and the respective current data; and
returning, by the data processing hardware, to the user device, the forecasted future data for each of the plurality of time series forecasts requested by the time series forecasting query.
2. The method of claim 1 , wherein the time series forecasting query comprises only a single query.
3. The method of claim 2 , wherein the single query comprises a Structured Query Language (SQL) query.
4. The method of claim 2 , wherein the single query comprises a reference to three columns of a table.
5. The method of claim 4 , wherein the three columns of the table comprise:
a first column comprising timestamp data;
a second column comprising the current data; and
a third column comprising time series forecast identification data.
6. The method of claim 1 , wherein each model of the plurality of models for each respective time series forecast comprises an auto regressive integrated moving average model.
7. The method of claim 6 , wherein training the plurality of models comprises performing hyper-parameter tuning.
8. The method of claim 1 , wherein forecasting the future data comprises modeling seasonal effects.
9. The method of claim 1 , wherein forecasting the future data comprises modeling holiday effects.
10. The method of claim 1 , wherein forecasting the future data comprises modeling drift.
11. The method of claim 1 , wherein forecasting the future data comprises performing anomaly detection.
12. The method of claim 1 , wherein selecting the one of the plurality of models that best fits the respective time series forecast comprises determining which model has a lowest Akaike information criterion (AIC).
13. A system comprising:
data processing hardware; and
memory hardware in communication with the data processing hardware, the memory hardware storing instructions that when executed on the data processing hardware cause the data processing hardware to perform operations comprising:
receiving a time series forecasting query from a user device, the time series forecasting query requesting the data processing hardware to perform a plurality of time series forecasts, each time series forecast being a forecast of future data based on respective current data;
for each time series forecast of the plurality of time series forecasts requested by the time series forecasting query, simultaneously:
training a plurality of models for the respective time series forecast decomposed into multiple components;
for each respective model of the plurality of models, estimating, using the respective current data, a respective relative quality of the respective model, the respective relative quality representing a quality of the respective model relative to each other model of the plurality of models;
selecting, using the respective relative quality of each respective model without relying on empirical means, one of the plurality of models that best fits the respective time series forecast; and
forecasting the future data based on the selected best fitting model and the respective current data; and
returning, to the user device, the forecasted future data for each of the plurality of time series forecasts requested by the time series forecasting query.
14. The system of claim 13 , wherein the time series forecasting query comprises only a single query.
15. The system of claim 14 , wherein the single query comprises a Structured Query Language (SQL) query.
16. The system of claim 14 , wherein the single query comprises a reference to three columns of a table.
17. The system of claim 16 , wherein the three columns of the table comprise:
a first column comprising timestamp data;
a second column comprising the current data; and
a third column comprising time series forecast identification data.
18. The system of claim 13 , wherein each model of the plurality of models for each respective time series forecast comprises an auto regressive integrated moving average model.
19. The system of claim 18 , wherein training the plurality of models comprises performing hyper-parameter tuning.
20. The system of claim 13 , wherein forecasting the future data comprises modeling seasonal effects.
21. The system of claim 13 , wherein forecasting the future data comprises modeling holiday effects.
22. The system of claim 13 , wherein forecasting the future data comprises modeling drift.
23. The system of claim 13 , wherein forecasting the future data comprises performing anomaly detection.
24. The system of claim 13 , wherein selecting the one of the plurality of models that best fits the respective time series comprises determining which model has a lowest Akaike information criterion (AIC).