In the era of artificial intelligence and big data, more and more cloud data and more and more intelligent models began to assist people to make various optimal decisions. From operational efficiency, cost saving, optimal configuration and other aspects, to achieve cost reduction and efficiency, further improve business efficiency.Jingdong, meituan, Didi, Shunfeng and many other well-known manufacturers have transformed their supply chain, intelligent order delivery, driver passenger matching, intelligent sorting and so on through the operation research optimization platform.
There are many links in the retail industry. In the supply chain from production to warehouse and offline stores, even if the demand for final products is very stable, the bullwhip effect often occurs. The reason is that when each node in the supply chain only makes production or supply decisions based on its neighboring demand information, the authenticity of demand information will go up the supply chain and enlarge step by step.More accurate demand forecasting is only a step of decision-making. There are also decision-making problems such as inventory ordering decision, price fluctuation decision and shortage game decision in the process of business change such as sales volume and process management.The bullwhip effect shows that even if the prediction is more accurate, if there is no effective management of the subsequent process decision-making process, the benefits of accurate prediction will be offset by the losses caused by unreasonable safety stock.
The decision-making process of many enterprises often relies too much on the personal experience of the corresponding positions. On the one hand, the information obtained by the employees is not complete, on the other hand, there is a lot of repeated work of estimation and comparison in the decision-making process, which leads to the low efficiency and instability of the output of the decision-making scheme. Repeated labor of employees limits personal growth, and the enterprise consumes human resources and valuable decision-making time.In view of the requirements of rapid replication, centralized and efficient decision-making, fast information feedback and prediction of planning decision-making effect of excellent planning decision-making experience methods of enterprises, singularity cloud has launched the application of decision engine on the basis of data in the middle platform.
Singularity cloud decision engine
Perfect data collection and management, information extraction of data, understanding the law of things, can not release the great value of data.In order to produce practical value, data must really improve the quality of decision-making and realize the automation, process and standardization of decision-making.
After completing the development of data middle platform for customers, it provides intelligent decision-making service based on data assets of middle platform. According to different scenarios, it selects decision-making methods such as maximum revenue expectation decision-making, maximum minimum revenue decision-making, minimum maximum regret value decision-making, Markov game decision-making, and solves decision-making objectives by combining operation research optimization algorithm and reinforcement learning.
In real life, there are many problems can be described as optimization problems, and then use the knowledge of operational research optimization to solve them.
The two core steps of comparison are modeling and solving.According to the mature software kits (CPLEX, gurobi, GLPK, lpsolve, SCIP…), the singularity cloud gives the baseline solution of the classic operation research optimization problem, and it can be put into trial operation quickly. In the process of operation, according to the core indicators of result evaluation, combined with operation research optimization algorithm and reinforcement learning, the algorithm and solution process are further optimized, so that the planning decision-making model, solution process and evaluation system can meet the planning decision-making process required by customer business development.
The singularity cloud prediction engine takes the demand prediction as the starting point, while the decision engine focuses on the efficiency and quality of decision-making in the process of implementation.For the seasonal impact of commodities and the instability of market supply, replenishment decision-making needs to be followed up reasonably; after the completion of distribution, replenishment from specific warehouse to store, and transfer from store to store still need a lot of work from the customer’s staff to generate replenishment and transfer schemes for each period; in order to complete the distribution, replenishment and transfer and ensure the market status at the same time We also need a reasonable distribution plan.
The core of planning decision is inventory allocation, including warehouse inventory, in transit inventory, store inventory, etc.Inventory management is to manage and control all kinds of goods, finished products and other resources in the whole process of production and operation of manufacturing or service industry, so as to keep its reserves at an economic and reasonable level. Using historical data to achieve real-time updated demand forecast, to provide replenishment suggestions for enterprises. Reasonable design of storage shelf placement, commodity area division, high and low shelf placement, optimal path allocation of warehousing and outbound, etc., can save huge costs and a lot of labor costs for enterprises. It can reduce capital occupation, improve inventory turnover rate, improve automatic management, improve personnel and equipment utilization rate, and reduce inventory burden.
Optimization of operation research to work out the optimal dispatching strategy
A big fashion customer of singularity cloud has thousands of offline stores and hundreds of SKUs in each store. Through historical data to predict the sales volume of each SKU in each store in the future, it is inevitable that some stores will be under stocked, while some stores will be under stocked. Then by transferring the goods of the overstocked stores to the under stocked stores, the overall gross of the company will be improved Profit. The logistics costs between stores are different, and the types of goods out of stock and overstocked are also different. Through the method of mixed integer programming in operation research planning, the optimal strategy of goods transfer is calculated. The model of mixed integer programming can be abstractly modeled as follows:
By modeling and solving the process of replenishment, the repetitive workload of the customer’s business personnel is reduced by 80%, and the planning decision-making time is shortened by three days.Business personnel can see more data basis when making decisions, and input and output of planning decisions are clear and efficient.
In the customer’s business process, a large number of links will involve decision-making issues,How to use data efficiently to drive decision is the core of singularity cloud decision engine.In the last column of startdt AI Lab, we mentioned the importance of accurate demand forecasting. However, in reality, there are always biases and uncertainties in forecasting, so it is necessary to make decisions under the multi-level uncertainties generated by different links. Combining demand forecasting and decision engine, make data decision more intelligent. In the future, we will continue to work in the field of demand forecasting and decision engine to help customers create greater value.