Gregory Kanevsky

Creating and plotting decision trees with the models in H2O

Machine and device generated data proliferated across industries and applications both vertically and horizontally. This effect is amplified by data sources linked with distributed networks and organized by intelligent processes. Businesses can't and won't afford to abandon these continuous and massive data streams and work hard on collecting, organizing and storing them in data lakes, data graveyards, and likes. We would like to present the unorthodox approach to processing, analyzing and… Show more

Machine and device generated data proliferated across industries and applications both vertically and horizontally. This effect is amplified by data sources linked with distributed networks and organized by intelligent processes. Businesses can't and won't afford to abandon these continuous and massive data streams and work hard on collecting, organizing and storing them in data lakes, data graveyards, and likes. We would like to present the unorthodox approach to processing, analyzing and extracting value from the stored data with the goal of improving and optimizing business processes. Show less

Архитектор решений Teradata Aster рассказывает о своем видении аналитики Больших Данных

We will show ways to use Aster and R on desktop to take advantage of the best features for both products and how seamlessly they integrate and compliment each other.

Optimizing advertisements is a daily challenge for TV and media operators. Existing approaches lack in effectiveness due to missing
and/or incomplete data, sampling errors, limited access to customers, their behavior patterns and transactions, and revenue attribution. The presented approach is a consolidated view of the systems involved in
advertisement for certain products: marketing, receiver device logs,
customer base demographics,… Show more

Optimizing advertisements is a daily challenge for TV and media operators. Existing approaches lack in effectiveness due to missing
and/or incomplete data, sampling errors, limited access to customers, their behavior patterns and transactions, and revenue attribution. The presented approach is a consolidated view of the systems involved in
advertisement for certain products: marketing, receiver device logs,
customer base demographics, and revenue transactions. By utilizing
Teradata Aster MapReduce Platform for data source integration, iterative
data discovery, time series generation, performance metrics calculation,
we devise and execute a predictive model to select ads based on
performance metrics. Teradata Aster’s ability to process, aggregate and
perform analytics on the large volume and variety of multi-structured
data achieves classification models that are rich in their predictive power
and estimated effectiveness gain of double digit percentages in lost
advertisement revenue. Show less

Background image credit