Anatol Gaina

Freistellung - ready for new challenges

Clustering of claims as mixture of numerical and textual information in Spark / Azure environment for correct repair cost estimation. Replacing of complex models with a family of more simple models, built on homogeneous clusters of data. Optimized data models by implementing machine learning algorithms and advanced statistical techniques - family of time series analysis models for better prediction of incoming tickets volume. Parts recommendation system for car repair workshops.

Automation of audit process. Text analytics of different text corpora. Extraction of most relevant information for new audit study from audit manuals and resolved aud it cases. Automatic key-phrase extraction, using extremes of sequence of TF / IDF scores of words as base for segmentation Fuzzy clustering of words, key-phrases and text chunks Splitting of documents in maximal homogeneous chunks Detection of topics in text chunks Clustering of documents by multi-level association / dissociation

Dream job for me should satisfy next conditions: 1. Interesting data - heterogeneous numerical data and textual labels. This combination allows me to find interesting patterns and insights and add an interpretation in business terms - customers, products, segments etc. 2. Strong Data Science Team - only in condition of tough competition my performance is highest. 3. Understanding of real value of big data by management.

Analysis of flights data in aerospace industry (Airbus Civil). Substitution of hard calculations, performed by flight engineers with probabilistic high accuracy model. The goal - to speed up the development of new aircraft. Clustering of load-stress data like moments and forces for different wings elements in different flight stages. Analysis of structure capability envelope, using pipeline clustering / decision tree / multiple linear regression

Creation of concepts of visualization of big data. Creation of specifications for General Scheme of Clustering. Implementation of Malt Clustering algorithm in Spark. Creation of architecture of Big Data Mining Portal. Implementation of my original algorithm of attributes' estimating by contribution in clustering. One-way clustering. Kernels' detection in big data.

Retail - customers and product segmentation, detection of patterns in buyers behavior, estimation of consumers preferences for the elaboration of bonuses and discounts policies in loyalty program, solving of basket problem. Banking - fraud detection software using supervised and unsupervised learning. Risk management software by clustering of credits information. IT - development of Real Application framework.

Pattern recognition in biological data

Calculus. Wiener-Hopf equation. "Some convolution equations, generated by group of operators"