In this chapter, a forecasting approach based on the public’s expectations is proposed. With wide use of the internet, web search data can reflect the public’s attention and expectations. First, data extraction. By using different keywords, the Google search volume index (GSVI) and Baidu search volume index (BSVI) are extracted from Google Trends and the Baidu Index, respectively. These two indices are used to represent the public’s expectations. Secondly, data selection. The Granger causality test and grey relational analysis are utilized to analyze the dynamic relationship between the RMB exchange rate and the web search data, including GSVI and BSVI. Thirdly, data computing. The kernel extreme learning machine (KELM) forecasting approach based on web search data to present the public’s expectations is proposed, by integrating GSVI, BSVI and KELM to forecast the daily and monthly central parity of the RMB against the US dollar. And multiple evaluation criteria are employed to comprehensively evaluate the forecasting performance of the proposed approach and benchmarks.