The invention relates to the technical field of
big data analysis and supply chain management, and discloses a fast moving
consumer goods sales prediction and
inventory optimization method based on
big data analysis, and the method comprises the following steps: S1, obtaining a historical sales
record of fast moving
consumer goods, a
sales promotion discount, a
network media impact factor, and a traffic busy index of a position where a store is located; s2, generating a unified
feature vector Xt by using the extracted multiple pieces of information; s3, constructing a fast moving
consumer goods prediction sales model FMCG, and introducing an attention mechanism and a bidirectional LSTM model BiLSTM into the model; s4, constructing a dynamic safe inventory
function model P; and S5, model optimization: performing optimization by using a
loss function in combination with an Adam optimizer. The method solves the defects that in the prior art, the data island problem is obvious, sales, inventory and external factor data cannot be fused, the model generalization ability is insufficient, and the method cannot adapt to the fast volatility of the fast moving consumer goods market. And feedback
closed loop is carried out, so that the prediction accuracy is improved, and the cost of fast moving consumer goods merchants is reduced.