5 questions to ask before starting a big data project
With regard to equipment performance, more data in the weekly reports of suppliers' key performance indicators (KPIs) and inventory levels may be the data that supply chain managers are most reluctant to deal with.
Big data 00
However, more data will appear every day: according to IBM's survey, the world generates 2.5 Ebyo bytes of data every day (that is, 1 billion GB). But this is not the case. According to IBM's calculation, 90% of the world's data has only been created in the past two years. The report shows that business use data can save millions of dollars and improve work efficiency in an unprecedented way.
When the enterprise budget is tight, it is not surprising that managers use a lot of data to improve efficiency. After all, many companies have spent more than ten years introducing or upgrading data processing systems and using cloud computing and/or networks. Now, supply chain managers are required to use these data, so it is easier to say.
People need to understand the challenge of blindly starting projects. Suresh Acharya, head of JDA Software Lab, analyzed the application of big data.
Acharia said: "Nothing is frustrating, there is a way to do it. He pointed out that the supply chain manager must ask himself five questions before starting a new project:
What is your business case?
Perhaps when managers try to apply data, their biggest problem is that no one can solve this problem. When starting a new project, the supply chain manager should have a specific business problem to solve (for example, excessive inventory), and be able to quantify (how much to reduce cost by 5%).
Acharya said, "If you understand the business problem you need to solve from the data, it is really the reverse." "What you want to say is: this is the problem I want to solve, it is my data, so it will collect or purchase and subscribe to help solve this problem?"
"So, you need to make sure you have a business case and try to solve a business problem," he added.
2. Is there a correct data source?
Considering that the big data project is a problem to be solved rather than a project to be completed, it may indicate that the currently available data is not the information needed to solve the problem.
"If you want to check whether the library keeps or is out of stock, do you have inventory data? Do you have sales points or orders? Or anything that may be data. You should adjust the business and data source you want to solve," Acharia said.
Asking this question may help determine what additional data must be collected before proceeding with the project. Additional product information from suppliers or different points of sale information from retailers may be required. If the partner is unable to provide this information, a new approach may be required.
3. Is your data available?
Similar to the second question, supply chain managers must be able to consider how to record and store data that can be used to solve business cases.
There are several types of data, but whether the data is structured or unstructured, endogenous or external can be distinguished based on the nature of the data item. In other words, retailers and manufacturers can collect various unstructured data, such as customer reviews of products. However, all parties collect, quantify and analyze data in different ways, so it may be completely different according to business needs. The ability to collect data sets cannot make them available; The parameters of the business instance determine whether it is available.
"If you think there is a lot of data, but don't check whether it adds or helps solve business problems, then you need to take a step back to find a solution," Acharya said.
Does the algorithm exist?
Once the business case is identified and the available data is considered relevant and useful, the enterprise must ensure that the problem can be solved based on the currently available algorithms If not, it is better to find a better solution.
"This is not to say that there is a method that can help you solve problems. As long as there are data and some problems are emerging academic circles or industries, these problems have not been solved," Acharia said.
"There may be a way to solve this problem, but this specific problem may not be solved, so you need to be ready to explore the algorithm," he added.
What is a sample?
If all the above conditions are met, the big data item is feasible. However, just because it can be achieved does not mean that implementation should be desired.
"All these should be tested on a small sample first," Acharia said. "If you want to try a small sample, then if it is feasible and has solved the problem, you can start to expand the scale“
Just as production runs need prototypes and samples, large data projects need to be tested to determine feasibility. Ideally, such a project will produce running results, but if there is a problem with the algorithm or implementation method, the results may be the wrong solution. For people, it is better to implement the project rather than implement it wrongly.
