부산대학교 도서관

Since the founding of our country's meteorological system has been very complete, the China Meteorological Administration released in 2015 after making 27 orders meteorological data towards open data (Open Data) a new stage, with the public sector can use meteorological information massive help enterprises, industries and mass data meteorological data has not been fully applied. This paper studies the impact on sales meteorological data, weather data and then use feature complete weather-driven industry sales forecast. We have two retail industry sales data, for example, combined with weather information provided by the Bureau of Meteorology analysis. We joined the economic factors in the analysis, as the Shanghai Composite Index and CPI data to provide external environment support. Different from traditional forecasting, meteorological data, we not only know the information at present point in time, there are at present readily available public accurate weather forecast for the next seven days. We use the popular machine learning algorithms to model the random forest, the result has been very good generalization. Our forecast model can solve the traditional retail industry sales forecast by practitioners of the limitations of subjective judgment, the use of big data analysis to achieve a more accurate and reliable guidance.