Dr. Kimin Oh, Senior Data Scientist, Samsung, will discuss demand forecasting using machine learning algorithms with a case study of Samsung Electronics. This is the major underlying theme for his presentation – at the upcoming DATAx Conference, taking place on November 6-7 in New York.
Dr. Oh recognizes that there are many difficult problems in demand forecasting, such as shifting seasonality, cold start situation, promotional spikes and so on. He will discuss the attempts to solve these problems and will share Samsung’s strategies including failures and success stories. Attendees will be able to take away some examples and insights from his session.
Dr. Oh recently identified a challenge with the quantity/quality of data when we try to utilize it. Stating “we usually meet the limitations to using a given data, because data has not been collected from the point of view of data analysis. That’s why it’s often difficult to get satisfactory results in many cases, although there are data and problems to solve.”
He went onto acknowledge how important it is to get your data, consider what you want to do and what you can do with it from the time of collection and creation.
When asked about where data science is heading Dr. Oh identified that there is a growing demand from companies to solve various problems by applying data analysis methods. Customers who request data analysis want to get the best results among all possible methods. For that reason, many companies are developing tools, platforms, and services for automating data analysis. In this process, not only the general methods but also new domain-specific methods are being developed.
Dr. Oh offered up some great advice for recruiting and maintaining top talent and for leaders looking to build high-performance teams.
“It is important to recruit people with the best skills. However, I think the most important thing is the organization of the team members with a proper structure. In my opinion, the key is to determine the job-related members, the appropriate number of people and the effective structure of these members. The overall data analysis process is divided into different types of work, so people in various roles are needed such as data engineers, data scientists, data analysts, and analytics managers. Finally, it is important to design a team structure so that these members can do their job effectively.”
When asked about his current Artificial Intelligence (AI) or Machine Learning (ML) initiatives, Dr. Oh shared his interest in the generative model for data augmentation. The data augmentation method has been used as a method to overcome the lack of data quantity and avoid overfitting. This method has been mainly applied to the field of image classification. In recent years, there have been many cases of using Generative Adversarial Networks (GAN) in tabular data and time-series data, but most of them are focused on privacy and classification problems. So, his team is trying to develop generative models for time-series and tabular data.
To hear more from Dr. Kimin Oh, reserve your place at the upcoming DATAx New York Conference on November 6 -7. Act fast to secure your ticket at the discounted early-bird price. [LEARN MORE AND REGISTER NOW]
Visit Argyle Executive Forum's 2019 CIO Think Tank: How to Optimize Multiple Cloud Platforms in New York, NY on Sep 26, 2019