Abstract: Data is becoming an integral part of digital marketing, as businesses realize the power of information to create successful campaigns and see real-time results.
In 2017, big data continues its growth as an important part of supporting business decisions. Armed with information on customer behaviors and purchases, we are now able to build profiles or user personas which ensure each marketing effort is geared toward a specific type of customer.
However, the crucial part of business reasoning is to examine large data sets to uncover hidden patterns, unknown correlations, market trends, customer preferences and other useful business information. This step is usually connected with looking for new tools, libraries and methods which performance is good enough to deal with large amounts of data.
This presentation shows how to combine efficient data processing on Amazon Redshift with deep exploration of behavioral Internet data using machine learning algorithms like k-means, Self Organizing Maps, Auto Encoders, Random Forests and Deep Learning.
To gain useful knowledge from raw data about user behavior on our websites and mobile apps, we need to combine efficient data processing and data wrangling tools with deep analytical insights. It turns out that applying easily scalable cloud data warehousing tools like Amazon Redshift, we are able to preprocess massive data sets and prepare data inputs for machine learning algorithms in a very fast way. The analytical part of the reasoning process is a unique combination of unsupervised and supervised learning. We use unsupervised learning to obtain the general data overview and to define characteristic groups of users. The supervised part is to find more features that describe our model groups to extend their potential and make them valuable from the business point of view.
Bio: Pawel is a Senior Data Analyst at Naspers (a global internet and entertainment group which owns Olx Group, Letgo, Brainly, Showmax). He deals with billions of data logs coming from OLX Group services and turns them into value for business using advanced data processing technologies and machine learning algorithms.
He first faced predictive modeling problems at Aviva, where he was building a Predictive Analytics Team from the early steps. Then he joined Grupa Wirtualna Polska – one of the biggest internet publishers group in Poland, where he had to deal with massive data sets to provide business teams with valuable behavior patterns of the users. Currently, he is leading ad-tech optimization project at OLX Group.
He is the author of several publications concerning recommender systems and intelligent data processing and has been published in top ranked scientific journals. He has graduated from the International Computer Science PhD Programme in the Polish Academy of Sciences. Pawel is passionate about statistical reasoning and constantly curious about getting new insights from the data.