Toward Data, the New Oil: Success on Supply Chain Management

Data Engineering DynaSys Blog Diadie Sow

In a highly competitive market environment, companies must focus on product improvements to meet new customer satisfaction and remain competitive in their journey towards becoming market leaders.

Nowadays, digital is so important that companies receive large amounts of data every day, therefore, data is a key factor in industry improvement. “Data is the new oil”; this concept of “new oil” is often credited to Clive Humby, a British mathematician. Humby established the likeness between data and oil using the principal that even though data is more valuable, it needs to be processed and analyzed like oil needs refining before its true value can be seen.

Data Engineering

Analyzing and processing data allows one to store, visualize, and apply mathematics and statistics to the data in order for companies to gain insight. Once the data has been manipulated, one can infer probabilities or likelihoods that something will happen, more precisely, companies will be able to communicate further with data.

The more data companies have, the better they can capture relationships and inferences within the data; this is a key factor in today’s industrial market as we have a lot of data. However without big data analysis, there is a lot of wasted data, and analytics without data is just a mathematical and statistical tool. The key to success is lots of data analytics, and processing and computing to generate intelligence information. Furthermore, to exploit the massive data collected, industries need a highly qualified technician, i.e., a data analyst who is skilled in computer engineering and data processing tools that could turn raw data into useful information.

Despite companies having lots of data, the success of data engineering is made possible due to the availability of massive computing power which otherwise would not have been able to be processed or computed due to its volume. Massive computing power revolutionized the field of machine learning by allowing algorithms to be converted within a reasonable time, which was not the case before.

Machine Learning

Machine learning is the art and science of teaching computers to teach themselves using data. The most successful type of machine learning algorithms are the ones that automate a decision making processes by generalizing from known examples (couples of inputs-outputs) known as supervised learning algorithms. Unsupervised learning is another common algorithm used.. In unsupervised learning, only the input data is known and there is no known output data given to the algorithm; this later can find patterns and categorize the data.

Many applications of machine learning are centered around customer segmentation to understand the best marketing strategy, product recommendation, fraudulent transaction detection, image classification, speed recognition, and demand forecasting for each segment. These applications of machine learning are done by a data scientist. A data scientist is different from a data analyst in regards to needing a superior understanding of mathematical knowledge like linear algebra optimization to be able to understand and solve machine learning models.

Data Engineering With Supply Chain Management

“One of the main purposes of supply chain collaboration is to improve the forecast accuracy” (Raghunathan, 1999). When it comes to demand forecasting, machine learning can be very helpful, particularly in complex scenarios. Unlike traditional statistical forecasting methods, machine learning uses all of the available data information to compute the best forecasting results; trade promotions, media events, weather data, new product launches, social listening, and complex seasonality are examples of common machine learning forecasting.

For example, when selling used goods online, a combination of tiny, nuanced details in a product description can usually make a large impact on creating a buzz. Avito, Russia’s largest classified advertisements website, recently realized that sellers on their platform sometimes feel frustrated with both too little demand (indicating something is wrong with the product or the product listing) or too much demand (indicating a hot item with a good description was underpriced). Therefore, they launched a machine learning competition on Kaggle’s website (see www.kaggle.com) to predict demand for an online advertisement based on its full description (title, description, images, etc.), context (geographically where it was posted, similar ads already posted) and historical demand for similar ads in similar contexts. The results allowed them to gather better information to advise sellers on how to best optimize their listing and provided an indication of how much interest they should realistically expect to receive.

DynaSys Machine Learning Strategy

Today at DynaSys, we “co-experiment” machine learning clustering and predictive algorithms with a few of our customers in hopes of helping them improve their accuracy. We challenge the traditional demand and supply chain planning (DSCP) statistical forecasting to get better results. Also, we are planning to include these machine learning models in our cloud solutions, including supervised and unsupervised solutions. Since the power of machine learning is associated with big data qualified by data scientist, we believe that in the near future, we will be able to run our customer’s supply chain at peak efficiency.

Diadie Sow
Diadie has a master degree in applied mathematics and computer sciences. He also obtained a master of science in pure mathematics, a master 2 in applied mathematics with option "Optimization, game theory and modeling in economics "at Polytechnique X and a master in mathematical modeling with option "Optimization and operational research" at the university Paris 1-Panthéon Sorbonne. After that, Diadie got a PHD in computer engineering about "Prognosis of complex systems for proactive performance management" at l'Ecole des Mines at Alès. He started to work on Machine Learning there and is now continuing his work at QAD DynaSys, where he “co-experiment” machine learning clustering and predictive algorithms with customers in order to bring, in near future, new machine learning features to our DSCP solution.

LEAVE A REPLY

Please enter your comment!
Please enter your name here