DATA SCIENCE
We follow CRoss-Industry Standard Process for Data Mining.
Established in 1996, still relevant today.
Our R&D process iterates through 3 main parts: Analysis, Research and Production.
BUSINESS
UNDERSTANDING
Investigate your problems
Describe your goals
Determine metrics
DATA UNDERSTANDING
Collect data
Form hypotheses
Explore all datasets
INSIGHTS
The main results of this stage are meaningful insights for your business: statistics, trends, ideas, detected problems or bottlenecks in processes, KPI and metrics.
That is the basis for Research.
ANALYSIS
RESEARCH
DATA PREPARATION
Remove irrelevant data
Clusterization
Feature engineering
MODELING
Identify patterns
Train predictive model
Finetune model
EVALUATION
Test model
Measure success
Detect problems
MATH MODEL
Based on achieved performance on KPI we decide upon the next actions.
We can either continue Analysis & Research or proceed with Production.
PRODUCTION
DEPLOYMENT
Deployment plan
Delivery report
PRODUCT
Our work does not stop after the deployment. Usually, we continue the R&D process and update the product adding new features and functions, refining a previous model or solving a new task.