Everyone is looking to make use of AI. Companies with a good set of data can get the upper hand against  companies that will be using a common set of generic data from the web that all companies have access to. Maybe you’re not completely sure what you can do with AI, but one thing is for sure: if you’re going to use your own data, you need to get it ready. That requires AI
Data Preparation.

What does AI Data Preparation look like? Let’s sketch out a
few simple examples of data that is not ready for AI:

• Duplicate data in one spot

Example: two records of the same customer with slightly
different names

• Duplicate data in different spots (which one is the source?)

Example: Customer data in one table but duplicated in another table for a different system and there you have added some information about the customer

• Fields where the values have different patterns – like usernames which should have a specific format
• Unstructured data (in NoSQL) has varying structures (that is: different fields per record)
• Data in text and Excel files – not such a huge problem but better to have it in SQL
• Data labeling: what does a field mean? If you don’t know, how will AI know?
• Data up to date (which details are old and which ones are current)
• Data consolidation: data in different tables but could just as well be in one table with categorization
• Categorization of data is not hierarchical and or could be better categorized

When we solve these data issues and create real consistency, we provide AI systems a data set that they can more easily learn from.