Within 1 to 2 weeks our team conducts a preliminary analysis of available data sources, engages in conversations with customers to gain full business domain understanding, and proceeds to design a series of possible datascience projects and data governance recommendations to maximize the generation of predictable value from data within a given organization.
DATA SOURCE IDENTIFICATION
Understanding and documentation of data sources, methods of denormalization, possibilities of valuable data projects, and governance recommendations surrounding the initiative.
We help our clients team understand the scopes of possibilities within datascience projects and guide them to take the decisions that maximize the value to their organizations regarding their engagement with datascience. We introduce our clients to our rapid data auditing process.
DATA OPPORTUNITY PRIORITIZATION
We use design thinking methodologies to combine our newly gained knowledge regarding the sources of data, our background in science and engineering, as well our insights gained by conversations with out clients to create a set of datascience project implementations ranked by estimated effort as well as estimated business value impact.
DATA SCIENCE AS A SERVICE
Within 4 to 8 weeks, our multidisciplinary team of domain experts,engineers, scientists and developers will sprint through a data specific business problem and delivery high quality insights and visualizations.
Following core Agile principles, our goal is to deliver intermediate and final results as fast as possible to our client in order to continue adding value throughout the use of iteration.
Our process follows a 5 step plan
Data Problem Definition
We start by engaging with our clients by understanding all aspects of the business problems they are trying to slove, helping them define their problem in terms of data.
Data Sourcing identification
We identify and analyze the data sources avalible to the organization.
Data Extraction, Cleansing and Transformation
After initial identification our team engages in the transformation and extraction of data from original sources into usable formats.
Modeling and Analysis
Combining feature engineering, model training, and model evaluation methodologies our team builds a series of analysis that provide high quality predictable value to the domain they are applied to.Some techniques used: Applied Statistical, Machine Learning, Fuzzy Logic, Survival Analysis, Applied Graph Algorithms.
After our analisis and statistical modeling phases are complete, we proceed to productionize our work, allowing our clients to easily and intuitively consume our analysis. Some of the methods of deployment we use are: Model Storage, Web Services, Intelligent Applications, BI Dashboards, RealTime Web Dashboards, and many more.