A merchant passenger flow volume multi-factor analysis and prediction method based on sparse regression

Abstract

The invention relates to a sparse regression-based merchant passenger flow multi-factor analysis and prediction method. The method comprises the following steps of: 1, preprocessing historical passenger flow data; 2, constructing a merchant passenger flow volume multi-factor dictionary; and 3, solving and predicting the sparse coefficient. The multi-factor sparse regression prediction method hasthe beneficial effects that the multi-factor sparse regression prediction method provided by the invention is obviously better than other prediction methods of several comparisons; compared with a sparse regression model without additional factors, the prediction effect of the multi-factor sparse regression prediction method is improved by 4.86%; when the error is 0.2 and 0.3, compared with othermethods, the multi-factor sparse regression prediction method has 10%-50% improvement on the basis of the number of merchants. Training time and prediction time of the sparse regression model and themulti-factor sparse regression model are far less than those of other two models;the multi-factor sparse regression prediction model provided by the invention is more excellent in performance.

Description

technical field [0001] The invention relates to a method for forecasting business traffic, more specifically, it relates to a multi-factor analysis and prediction method for business traffic based on sparse regression. Background technique [0002] At present, traditional time series forecasting methods and time series forecasting methods based on machine learning have related applications in business traffic forecasting. Comparison of Two Time Series Models in Passenger Flow Prediction, Liu Jianjun et al., Computer Engineering and Application, in December 2014, announced the passenger flow forecast results of fuzzy time series model and seasonal model in a shopping mall in Nanjing, and found that the season considering data characteristics The model outperforms fuzzy time series forecasting methods. A big data forecasting method for business traffic flow based on time series decomposition, Wang Jin et al., invention patent, disclosed a business traffic forecast method base...

AI Technical Summary

Problems solved by technology

However, in the actual application process, the traffic flow of merchants in the future is closely related to many external factors, and is greatly affected by external factors. It is difficult to accurately model this problem only by relying on historical traffic data.

Images