Business intelligence basics

In this guide, we will see which functionnalities each type of analytic solution has to offer,
how they compare to each other and how to make the right choice for your needs.
Classical solutions for Business-Intelligence can be roughly categorized as followed:

Data integration: ETL

The main objective of ETL tools (Extract, Transform, Load) is to gather the content of various databases or operational systems across an organization and centralize all data in one place, named “data warehouse”.

Read more

Data warehousing

A data warehouse is a centralized place where you can find all the data of your company. Data Warehouse software are very often simple database: Oracle, Teradata, SQL Server, MySQL,…

Read more

Reporting & OLAP

Reporting software (i.e. OLAP tools – Online analytical processing tools) allow you to create charts and graphs about the evolution of different KPI (Key Performance Indexes) that characterizes your business.

Read more

CRM softwares

The CRM acronym stands for “Customer Relationship Management”. This category covers Operational CRM and Analytical CRM softwares.

Read more

Advanced Analytics: 2 different approaches

Analytical CRM can be classified in 2 categories: Analytical CRM tools based on segmentation techniques and Analytical CRM tools based on predictive techniques. By its very nature, segmentation is a technique well-adapted for exploratory work. In opposition, predictive analytics is discriminatory in nature. Analytical CRM tools based on predictive techniques usually generates a higher ROI for marketing campaigns.

Read more
segPredLifts
datasetForScoring

Of the utility of the test dataset

When comparing the accuracy of 2 different predictive models to know which predictive model is the best one (the one with the highest accuracy & the highest ROI), you must compare ONLY the “lift curves computed on the TEST dataset” and if possible, use always the same TEST dataset.

Read more

Not all lifts are born equal

When we are constructing a model that makes no mistakes on the “training dataset”, we obtain a model that doesn’t perform well on unseen data. The “generalization ability” of our model is poor. The predictive model is using some information there were in reality noises. This phenomenon is named “Over-fitting”. A model that “overfits” the data can have an accuracy of 100% when applied on the learning set.

Read more
predictiveModel2