3. Preparing data

It's an illusion to think that the data found in the various sources will be ready to use.

The analyst will have to apply several successive processes:

• control of aberrations and extremes with possible suppression ;

• data homogenization where possible (e.g., for numerical data to be converted or names to be standardized); see figure "Example of Open Refine's name homogenization processing". Open Refine is a tool that enables data to be reprocessed, standardized or deduplicated;

• data structuring, for example using segmentation solutions (Web scrapping) or extraction of named entities;

• data exploitation (statistical processing, spatialization,...