Research On Fmcg Demand Forecast Of Q Company Based On Machine Learning

The total sales volume of fast moving consumer goods(FMCG)increases year by year.People cannot live without FMCG,and the consumer market of FMCG is quite broad.Although the prospect of FMCG industry is very optimistic,the competition of fast moving consumer goods in the current market is very fierce,and the phenomenon of the rise and fall of enterprises in the industry often appears.Based on the characteristics of FMCG and this industry,in the fierce competition and changing market environment,companies need to meet the needs of consumers in the market,and at the same time,it also needs to enhance the core competitiveness to gain profits and ensure the long-term stable operation of companies.FMCG have strong substitutability,so their demand in the market is very unstable.In order to ensure that there is no shortage of products on the market and maintain the normal supply of products,enterprises tend to hoard a large amount of inventory,but this will lead to enterprises to bear huge inventory costs,limit the flow of cash flow,and ultimately reduce the profits of enterprises.Demand forecasting is an important joint in logistics management.Improving the accuracy of demand forecasting is of great significance for companies to reduce inventory and rationally arrange production plans.It also plays a non-negligible role in enhancing downstream customer satisfaction and enhancing the core competitiveness of companies.This paper takes Q company as the research object,analyzes the existing problems in demand forecasting of this company,and optimizes the management of demand forecasting methods.First of all,the current sales of FMCG products to downstream customers in Q company are described and analyzed,and the current problems of Q company are found and the root causes of the problems are explored.Then,based on the knowledge of product sales,planning and demand forecasting management in modern supply chains,the historical sales data of FMCG of Q company was collected and cleaned.According to the total sales and sales quantity of products,all FMCG products sold by Q company are divided into four categories,which are cleaning,nursing,sanitary products and special products.The traditional time series prediction method ARIMA,random forest,XGBoost and LSTM neural network models in machine learning were proposed for prediction.In addition,combined forecast model was used to forecast the demand of Q compare’s FMCG.And evaluation indexes such as MSE,MAE and MAPE of each model were compared to obtain the optimal prediction model for each type of product.Finally,the corresponding solutions are proposed for the existing demand forecasting problems of Q company.This paper has great application value to the optimization of demand forecasting management for consumer goods of Q Company.Adopting the demand forecasting model proposed in this paper can help Q company obtain an agile supply chain,reduce unnecessary inventory costs,improve order response speed,and make Q company gain more competitiveness in the Chinese market.At the same time,it provides some reference value for other enterprises in the demand forecasting of fastmoving consumer goods industry.