Data Science | Management Consulting

Data Science (DS) creates deep insights for the business. It enhances the traditional BI model by adding stochastics and machine learning. DS uses all sources of intelligence to predict the immediate future and recommend a course of action.

Business Intelligence

Business Intelligence (BI) provides actionable intelligence to the business.

Over the past 10 years more than 100.000 decision makers have been using BIPortal reports. Most clients use them to optimize their supply chain (SCM) or customer relationships (CRM).

Whereas reports rely on the analyst to interpret and integrate the infomation, a Recommender System (RS) automates this process such that analysts can focus on optimizing the overall performance of the decision-making system.

Well known examples are fraud detection, the reduction of customer churn, permission marketing, and demand management.

BIPortal™ helps clients developing RS based on Prescriptive Analytics.

Competitive Intelligence

Competitive Intelligence (CI) as defined by the Strategic and Competitive Intelligence Professionals (SCIP) uses all legally available information to give the business a competitve advantage.

BIPortal uses the competitve intelligence data from marketing research firms like ACNielsen, GfK, and Drotax.

In addition these processed data sources BIPortal collects data from news feeds and social media for social media analysis and monitoring, campaign tracking, employer branding, target audience analysis, crisis Identification, prevention, and management.

Often this raw data first needs to be transformed into structured data before it becomes useful. Therefore we use Natural Language Processing (NLP) techniques to disambiguate and extract information hidden in unstructured data.

Social Intelligence

Social Intelligence (SI) describes relationships amongst individuals and organizations.

BIPortal uses the following types of social intelligence:

Customer Loyalty Program: Still the primary source of personal information
Analytical CRM: Measure and act on the social fabric of your customers
Social Marketing Intelligence: Identify alpha users influencing your community through sentiment analysis
Employee Satisfaction and Performance: Improve your client-facing staff's ability to better communicate your product's advantages
Service quality: Obtain feedback through mystery shoppers
Customer interaction: Analyze call center records, customer emails and letters
Credit ratings: Optimize your debt management

Ambient Intelligence

Ambient Intelligence (AmI) makes available all information associated with a transaction. Marketeers need to know why person X bought product Y in our branch Z at time T and what else (basket B) could have been sold had it been offered.

Typically this information is sensor-generated like movement profiles of potential customers (via mobile phone traces like Google Location Reporting, car tracking systems like eCall, Telefonica m2m and Toll Collect, in-store customer tracking like Euclid, shopkick and Apple's iBeacon, credit card profiles, or click-stream analyses like Google Analytics).

Contents of mobile applications, HR, store layouts, product pressure, noise levels, music types, commercials, promotions, announcements, illumination, RAL codes, queue lengths, POS displays, temperature, weather forecasts, economic indicators, sentiments, consumption data, registration records, traffic data, river stages, events and TV program data can influence customer behavior and hence should be added.

A single transaction therefore can entail thousands of ambient data points multiplying the data volume of a traditional Data Warehouse (DWH) by a factor of 100-1000. The handling of these petabytes (PB) of data is now called 'Data Lake'.

Operational Intelligence

Operational Intelligence (OI) is used to describe optimal strategies to decision-making problems. The best solution can often be calculated using methods commonly called Operations Research (OR) and include models of stochastic processes like e.g. sales. OI therefore is an integral part of any Prescriptive Analytics solution.

BIPortal focuses on but is not limited to the following methods:

Decision Trees: Graphical decision-modeling for real-time decision-making
Multi-criteria Decision Analysis: Evaluation of multiple conflicting criteria in decision-making processes
Assignment Problems: Cost-optimal scheduling of tasks and resources
Inventory Management: Optimization of safety stocks to minimize costs
Master Data Management: Determination of similar products, customers, and branches for consolidations and like-for-like comparisons

Computational Intelligence

Computational Intelligence (CoI) uses nature-inspired methods to solve complex optimization problems. Most machine learning (ML) algorithms rely on CoI.

BIPortal uses the following types of computational intelligence:

Evolutionary Algorithms: EA have the advantage of being self-learning, i.e. as long as the problem is well defined it approximates the optimal solution in iterative steps
Neural Networks: NN mimic the brain by enforcing those connections that help identify a given pattern. Its optimal structure can be generated by an EA (see above)
Fuzzy Logic: FL allows for the modeling of language concepts as used by humans. Terms like 'hot' and 'warm' are usually not defined by specific value ranges but overlap and are subjective to the speaker. FL can therefore be used to train a NN (see above)
Bayesian Networks: BN allow for simple but effective statistical reasoning based on relationships among language concepts (see FL above). BN are therefore the first choice when it comes to automating low-level (mass) decision making with Prescriptive Analytics.