Data Warehouses and data mining to predict the future

The majority of enterprises, at least the competitive ones, have informatics systems of relational databases to maintain updated the daily racking of information. The multidimensional databases are different from the relationals, because store historic information. See it easily with an example: the buy and devolution of a product in a store would leave intact the quantity of stock of a product in the relational data base, however, in the multidimensional database, would have how have evolved the stock during the time, storing the exit and the entry of the product. So the data warehouses work with historic information and the relational databases allows that the enterprise can do their diary work.

The data warehouses can need several hundreds of GB to store all these information. Its design is very different compared to the relational databases, because with this change of design we save a lot of time at the time of query all that GB of data. The information is obtained from the relational databases, using the ETL processes (extraction, transformation an load), that are launched automatically at night when all the users of the database are sleeping.

Through OLAP tools that make queries to the data warehouse to extract results that, also, after can be treated with artificial intelligence algorithms and statistic techniques to obtain new information and make predictions. In this way we can get information as: What kind of client buy this product? What products are sell with others and which are?, What discount I have to apply to get the maximum profit with the sells?, what is the trend of the sells according to the epoch of the year and the kind of product?, etc.

In this way we can find strange relations that never we can imagine, as for instance that the people that buy diapers in the supermarkets also buy too beer, for that is normal that these products are put together strategically.

With total security these tools also are used in Facebook, Google, Hotmail and others, to make profiles of our personality, with the finality of launch campaigns of ads adapted to us. In fact some tools of data mining, use as source of information blogs and social networks as Twitter, Facebook, LinkedIn, etc.

The enterprises will end knowing more of us than ourselves.