Data Mining - datosclaros

home : methodologies : data mining

Methodologies

Data Mining

Data Mining utilizes different methodologies (automatic learning, statistical analysis, information theory, etc.) to reveal hidden behavioral patterns, trends, and relations in data.
Data Mining extracts and transforms data expressed in “gigabytes” into useful information that is required for decision making.

Rules of Association

Rules of association identify patterns in data, associations, correlations, and causal structures among sets of items in databases. These rules are used in basket analyses, cross marketing, catalogue design, etc. A practical example is the proposal to group product packages that are likely to sell as a combo. A simple rule of association could be:

Clients who purchase product “A” or “B” and are between ages 20 and 40 = >

Will buy product “C” (probability of 70%).

The algorithms to discover rules of association are capable of analyzing hundreds of parameters from millions of cases in just a few seconds time.

Classification Models

These models describe and distinguish classes or concepts in databases according to their profile or behavior. The process is done with a classification algorithm that examines data to identify the common patterns that allow the data to be grouped. In this way, individuals with the highest probability of responding to a specific stimulus can be identified, thus improving marketing campaigns, reducing costs and increasing efficiency.

Clustering

Clustering is a way to group observations in clusters, based on the similarities among the observations. Clusters are created without the need to pre-define them; they represent the information discovered automatically by the algorithm. This methodology is utilized for tasks in which there is no classification information. One of its uses includes classifying clients according to their buying preferences.

Web-farming

Web farming is based on systematic use of a content present on the Internet in order to keep a company informed about topics essential to its business. The Internet is increasingly used as a source for information. Web-farming identifies information relevant to the firm’s business, connecting this information to the firm’s database and delivering the processed information to the people or departments indicated by the organization.

Website Data Mining

Automatic analysis of website visits can provide extremely valuable information. This kind of analysis offers much more than a simple count of hits. Website data mining can help clients identify the profiles and preferences of visitors by utilizing behavioral models (page jumps, amount of time spent in each page or entry and exit pages, etc.). Website Data Mining offers information on the products attractiveness and suggests better ways of presenting services to satisfy a client’s needs.