Research On Medicine Sales Forecast Based On Environmental Factors

With the rapid development of the pharmaceutical industry and the growing competition in the industry,enterprises are generally facing the pressure of high inventory costs,more expired drugs,long delivery cycle,and so on.Therefore,how to achieve zero inventory and minimize inventory costs has become an important research topic for researchers.Accurate sales forecasting is the key to achieve zero inventory.It is urgent to have a set of effective drug sales forecasting methods to provide scientific data support for the realization of zero inventory.This paper mainly studies the forecasting method of pharmaceutical sales.The real sales data of a pharmaceutical enterprise from January 2017 to October 2018 were taken as the research object.On the basis of in-depth analysis of data,the theory and application fields of various forecasting algorithms are studied.Combining with environmental factors(meteorological and air quality factors),a pharmaceutical sales forecasting model based on Xgboost algorithm is constructed.The main research work is as follows:(1)Chinese patent medicines and traditional medicines are classified into two categories according to drug categories,Considering the characteristics of clinical combined drug use,the association rules of drugs are mined by using FP-growth algorithm,finding frequent itemsets between drugs,and grouping them to determine the predictors.To a certain extent,this method solves the problem of high computational complexity caused by the variety of drugs,and at the same time,it is beneficial for enterprises to carry out drug combination sales,shelf placement,and optimize the inventory space.(2)According to the attributes of drugs and environmental factors,the impact of historical sales and environmental changes on future sales is studied.Based on the relationship between environmental change,human health and drug use,effective features are constructed by combining meteorological and air quality indicators;Considering the incubation period of the disease,the correlation and significance between sales volume and meteorological and air quality indicators in the same period and time lag period are analyzed,based on which the derivative characteristics are constructed.On this basis,feature extraction and feature selection are carried out to generate high-dimensional training set and further optimize feature engineering.(3)Based on Xgboost algorithm and the characteristics of environmental factors,a pharmaceutical sales forecasting model is constructed.Two contrast algorithms are set up,and the experimental results are analyzed to illustrate the impact of environmental factors on sales and the validity of the model.Experiments show that,the model has high prediction effect and good expansibility.It can be applied not only to drug sales forecasting,but also to other commodity sales forecasting related to sales volume and environmental factors.It plays a positive role in the research of sales forecasting methods.