Companies that want to be more competitive need to use the newest IT solutions that give them advantage on the market. It is not enough anymore to use standard IT tools such as BI or OLAP which provide analysts with information what has happened, now it is necessary to predict what will happen in the future.

In different business areas different questions are asked such as credit risk in finance or churn in telecom companies, while questions such as who will buy my product or service is same in all industries. With prediction analysis you can discover patterns hidden in massive data volumes, discover new insights, make predictions and immediately transform raw data to actionable insights.

Multicom’s expertise in business intelligence and data warehousing is empowered with prediction analysis skills that are more and more needed and used in today’s world of big data.

Business problems we are dealing with

  • Anticipating cross/up-sell opportunities
  • Anticipating resource demands
  • Improving marketing campaign response rates
  • Identifying customers likely to churn
  • Analyzing “market baskets” to discover associations, patterns and relationships
  • Reducing fraud
  • Customer life time value prediction
  • Anticipating future product demand
  • Credit risk prediction
  • SAS to Oracle Advanced Analytics migrations

How we do it

Multicom uses the cross industry standard processing methodology for data mining (CRISP-DM) which is proven for building predictive analytics models across the enterprise. It is an application neutral model that was devel- oped to standardise the industry and provide companies with a roadmap to more successful data mining by encour- aging best-practices. The process illustrated in the diagram is cyclical, meaning that creating a data mining model is a dynamic and iterative process.

After you explore the data, you may find that the data is insufficient to create the appropriate mining models, and that you therefore have to look for more data. Alternatively, you may build sever- al models and then realize that the models do not adequately answer the problem you defined, and that you therefore must redefine the problem. You may have to update the models after they have been deployed because more data has become available. Each step in the process might need to be repeated many times in order to create a good model.

Multicom support in a data mining process

Source system analysis

  • Data analysis
  • ETL/ELT data transformatio
  • Data cleansing
  • Programming support for data quality (Multicom mDQ)
  • Model build, validation and cross-validation
  • Program support for call-center and campaign-management