Research On Influencing Factors And Forecast Of Drug Sales In Retail Pharmacies Based On Machine Learning

With the development of the reform which aimed to separate the dispensing of medication from the provision of prescriptions in China,pharmacies gradually came to play a greater role.This paper mainly studied the influencing factors of drug sales in retail pharmacies and the method of forecasting sales.The influencing factors of drug sales were divided into three categories: internal factors,external factors and macro backgrounds in this paper,and this paper predicted drug sales based on the analysis of influencing factors.The main body of this paper was divided into three parts: the impact of internal factors and month on drug sales,the impact of retail pharmacy location,that is,external factors on drug sales,and the forecast of drug sales with limited-time data.In the first part,this paper used multivariate linear regression and seasonal index to analyze the influence of internal factors and month on retail pharmacy drug sales,and sales data for a two-year period related to AB Company's 101 retail pharmacies were analyzed.The results show that the number of drug varieties positively affects drug sales,and drug stores should appropriately enrich the number of drug types on the basis of considering cost and comfort caused by drug display.Price and price change negatively affect drug sales,and drug stores should consider the price elasticity before they raise price.The number of staff positively affects drug sales,and pharmacies should rationally arrange the number of service personnel while improving service quality.The availability of Medicare negatively affects drug sales,and pharmacies should eliminate the negative impact of medical insurance through improving the customer experience.The area of pharmacies positively affects the sales of drugs,and pharmacies should reasonably determine the area of pharmacies in consideration of cost,and optimize the decoration and furnishing of pharmacies to provide better experience at the same time.Season affects drug sales,and pharmacies should use the seasonal index to understand drug sales in different months.In the second part,by controlling variables employing the weighted Partitioning Around Medoids clustering algorithm and paired-samples t-tests,this study aimed to examine the impact of retail pharmacy location,that is,external factors on drug sales volume.Sales data for a two-year period related to AB Company's 101 retail pharmacies were analyzed.The results show that while considering that internal factors and macro backgrounds remain similar,the general order of the pharmacies' drug sales is that pharmacy located adjacent to the hospital < pharmacy located in the hospital < pharmacy located in the community,pharmacy located adjacent to the hospital < pharmacy located in the town(or pharmacy located in the business district).The findings reveal that retail pharmacies located in the hospital was promising due to the reform of the medical system,and retail pharmacies located in the community achieved good sales results,but the performance of pharmacies adjacent to hospitals was poor.This part proposes a method to study the influence of a single independent variable on the dependent variable in complex environment,and findings in this part can help drug companies in their decision-making when selecting the location of pharmacies.The third part studied the forecast of drug sales based on the analysis of influencing factors of drug sales.For newly opened retail pharmacies and new drugs,it is difficult to conduct time series analysis using ARIMA model due to the lack of long-term sales data.It is also difficult to carry out causal prediction because it is hard to take all influencing factors into account.In order to solve these problems,this part proposes a compound forecasting model for drug sales.The model consists of three parts.The first part is to obtain sales trend via exponential smoothing.The second part is to reduce redundant information through principal component analysis by category.The third part is to forecast sales applying backward propagation feedforward neural network.This paper predicts half-year drug sales of a kind of cold medicines in 24 pharmacies,a kind of stomach medicines in 51 pharmacies and a kind of antihypertensive medicines in 49 pharmacies according to two-and-a-half-year sales data of AB Company on a monthly basis,respectively.The results show that the correlation coefficients between the predicted values and real values are 0.77,0.84,and 0.85,and the standardized mean square errors are 0.48,0.37,and 0.30.This indicates that it is an effective model.On the one hand,this model will help retail pharmacies to formulate appropriate inventory,on the other hand,it also help other node companies within the pharmaceutical supply chain to optimize their own decisions accordingly.In this paper,there are 9 figures,8 tables and 66 references.