The sad truth that still dominates the world of analytics is the famous principle “garbage in – garbage out”. The seemingly large technology advancements in the field of Big Data are diminished by the low maturity of data governance in most organizations.
With our sensemaking services we help to bridge this maturity gap and enable organizations to be ready for advanced analytics.
- Data Collection When creating an analytical solution, the first thing to be considered is an observation space. In most cases, the initial observation space is not enough to reach the objectives of the analytics. We help to widen and prepare the observation space by understanding and qualifying the data sources, connecting to them using various methods (crawlers, APIs, indexing, CDC etc) and collecting the required data for further processing including initial format transformations. We are also experts in collecting web-based data through a massive web-crawler infrastructure and enabling connectivity to major social networks.
- Data Preparation
It is probably the most common delusion that analytics can run on unstructured and uncertain data. In fact, it cannot. All core analytics methods need fully structured data to be executed. Hence, you always have to prepare the collected data beforehand. We help to prepare the collected data of various types for analysis by developing feature and entity extraction components (e.g. text annotators or image object recognition), data normalization, cleansing and standardization methods and enabling organizations to create their own.
- Entity Resolution
When the collected data is structured and normalized, it is surprisingly not yet ready to gain insight from it. The data exists in the form of entities and relationships, where multiple entities may be duplicates of one same object and lots of ambiguities exist. Spotting the “True Truth”, a singular exclusive truth about entities and their relationships, is an absolute prerequisite for actionable intelligence. It is achieved through a series of techniques called entity resolution.
We offer services to our customer where we develop entity resolution pipelines as well as specific methods and algorithms for entity matching and deduplication, relationship scoring and other techniques to establish a singular truth domain for actionable analytics.
- Analysis and Prediction
This is the breathtaking moment – the jigsaw puzzle is completed, the Rubic’s cube is assembled. So it is time to actually understand, what is displayed on the picture, make decisions, take actions and learn. It is enabled with a rich set of toolkits and algorithms starting from traditional and ad hoc reporting to most recent adaptive and deep learning algorithms. We provide to our clients a range of services to create a working decision support solution following the industry standard CRISP-DM approach. Our team of data scientists and business analysts helps to build the best-fit model to solve industry-specific analytical tasks.