48 Pages Posted: 18 Dec 2018 Last revised: 17 Sep 2019
MacroXStudio; Harvard Business School; Microsoft Corporation – Microsoft Research – Redmond
Date Written: November 20, 2018
Traditional macroeconomic data used by economic agents to make decisions are noisy, lack richness, and produced with considerable lag. This chapter explores how alternative, web-scale data sources (“Big Data”) can help. We present a case study using a common alternative data source- web search to predict one of the most important data releases- non-farm payrolls (NFP). We discuss the efficacy of various machine learning (ML) techniques, the live performance of alternative data prediction models and the typical problems faced in practice.
Keywords: Big Data, Macro Forecasting, Alternative Data, Non Farm Payrolls, Artifical Intelligence, Machine Learning