Correcting Forecasts with Multifactor Neural Attention

Matthew Riemer, Aditya Vempaty, Flavio Calmon, Fenno Heath, Richard Hull, Elham Khabiri
Proceedings of The 33rd International Conference on Machine Learning, PMLR 48:3010-3019, 2016.

Abstract

Automatic forecasting of time series data is a challenging problem in many industries. Current forecast models adopted by businesses do not provide adequate means for including data representing external factors that may have a significant impact on the time series, such as weather, national events, local events, social media trends, promotions, etc. This paper introduces a novel neural network attention mechanism that naturally incorporates data from multiple external sources without the feature engineering needed to get other techniques to work. We demonstrate empirically that the proposed model achieves superior performance for predicting the demand of 20 commodities across 107 stores of one of America’s largest retailers when compared to other baseline models, including neural networks, linear models, certain kernel methods, Bayesian regression, and decision trees. Our method ultimately accounts for a 23.9% relative improvement as a result of the incorporation of external data sources, and provides an unprecedented level of descriptive ability for a neural network forecasting model.

Cite this Paper

BibTeX
@InProceedings{pmlr-v48-riemer16, title = {Correcting Forecasts with Multifactor Neural Attention}, author = {Riemer, Matthew and Vempaty, Aditya and Calmon, Flavio and Heath, Fenno and Hull, Richard and Khabiri, Elham}, booktitle = {Proceedings of The 33rd International Conference on Machine Learning}, pages = {3010--3019}, year = {2016}, editor = {Balcan, Maria Florina and Weinberger, Kilian Q.}, volume = {48}, series = {Proceedings of Machine Learning Research}, address = {New York, New York, USA}, month = {20--22 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v48/riemer16.pdf}, url = {https://proceedings.mlr.press/v48/riemer16.html}, abstract = {Automatic forecasting of time series data is a challenging problem in many industries. Current forecast models adopted by businesses do not provide adequate means for including data representing external factors that may have a significant impact on the time series, such as weather, national events, local events, social media trends, promotions, etc. This paper introduces a novel neural network attention mechanism that naturally incorporates data from multiple external sources without the feature engineering needed to get other techniques to work. We demonstrate empirically that the proposed model achieves superior performance for predicting the demand of 20 commodities across 107 stores of one of America’s largest retailers when compared to other baseline models, including neural networks, linear models, certain kernel methods, Bayesian regression, and decision trees. Our method ultimately accounts for a 23.9% relative improvement as a result of the incorporation of external data sources, and provides an unprecedented level of descriptive ability for a neural network forecasting model.} }
Endnote
%0 Conference Paper %T Correcting Forecasts with Multifactor Neural Attention %A Matthew Riemer %A Aditya Vempaty %A Flavio Calmon %A Fenno Heath %A Richard Hull %A Elham Khabiri %B Proceedings of The 33rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2016 %E Maria Florina Balcan %E Kilian Q. Weinberger %F pmlr-v48-riemer16 %I PMLR %P 3010--3019 %U https://proceedings.mlr.press/v48/riemer16.html %V 48 %X Automatic forecasting of time series data is a challenging problem in many industries. Current forecast models adopted by businesses do not provide adequate means for including data representing external factors that may have a significant impact on the time series, such as weather, national events, local events, social media trends, promotions, etc. This paper introduces a novel neural network attention mechanism that naturally incorporates data from multiple external sources without the feature engineering needed to get other techniques to work. We demonstrate empirically that the proposed model achieves superior performance for predicting the demand of 20 commodities across 107 stores of one of America’s largest retailers when compared to other baseline models, including neural networks, linear models, certain kernel methods, Bayesian regression, and decision trees. Our method ultimately accounts for a 23.9% relative improvement as a result of the incorporation of external data sources, and provides an unprecedented level of descriptive ability for a neural network forecasting model.
RIS
TY - CPAPER TI - Correcting Forecasts with Multifactor Neural Attention AU - Matthew Riemer AU - Aditya Vempaty AU - Flavio Calmon AU - Fenno Heath AU - Richard Hull AU - Elham Khabiri BT - Proceedings of The 33rd International Conference on Machine Learning DA - 2016/06/11 ED - Maria Florina Balcan ED - Kilian Q. Weinberger ID - pmlr-v48-riemer16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 48 SP - 3010 EP - 3019 L1 - http://proceedings.mlr.press/v48/riemer16.pdf UR - https://proceedings.mlr.press/v48/riemer16.html AB - Automatic forecasting of time series data is a challenging problem in many industries. Current forecast models adopted by businesses do not provide adequate means for including data representing external factors that may have a significant impact on the time series, such as weather, national events, local events, social media trends, promotions, etc. This paper introduces a novel neural network attention mechanism that naturally incorporates data from multiple external sources without the feature engineering needed to get other techniques to work. We demonstrate empirically that the proposed model achieves superior performance for predicting the demand of 20 commodities across 107 stores of one of America’s largest retailers when compared to other baseline models, including neural networks, linear models, certain kernel methods, Bayesian regression, and decision trees. Our method ultimately accounts for a 23.9% relative improvement as a result of the incorporation of external data sources, and provides an unprecedented level of descriptive ability for a neural network forecasting model. ER -
APA
Riemer, M., Vempaty, A., Calmon, F., Heath, F., Hull, R. & Khabiri, E.. (2016). Correcting Forecasts with Multifactor Neural Attention. Proceedings of The 33rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 48:3010-3019 Available from https://proceedings.mlr.press/v48/riemer16.html.

Related Material