Académique Documents
Professionnel Documents
Culture Documents
Means
collect
Tools
Consolidate
Methods
Modeling restitute
Selecting the data Sort, associating or separating Data Developing summary operations Presenting the results in summary form
DM introduces an extra dimension which is "exploratory" modeling, that transform Data to Knowledge
User
Decision Maker
Data Analyzer
DB Administrator
Processes Processes
Defining the data mining problem
Collecting data Detecting the data
Processes Processes
Defining the DM problem
Expertise
Processes Processes
Collecting data
We determine the general structure of data and the rules used . The selection of data must be optimal and may require consultation of experts, to determine attributes which describe the problem.
Processes Processes
Detecting the data
We must build an information base which allow learning, it means the construction of models by looking at past similar events.
Processes Processes
Correcting the data
Perform a potential quality diagnostic of data in order to enhance results and minimize anomalies.
Processes Processes
Variables
Transforming variables in order to prepare them for the analytic work. intervening on variables better exploition by modeling tools.
Processes
Research of the model
Extract data from a volume of blurred data, then present it in a summary form
Processes
Results evaluation
Evaluation of results give an estimation of the model quality , i e, its capacity to determinate new correct values; qualitative evaluation : Illustrate the influence of a factor quantitative evaluation: make future information more reliable.
Processes
Knowledge Integration Knowledge is useless unless it is converted into a decision and then to an action. It is essential to implement the model and its results in computer systems or in the processes of the company.
TECHNIQUES
Techniques used according to their <Origin>
Statistics Theory of estimation, econometric tests Maximum likelihood and squares regression
Data analysis (exploratory statistics) Factorial description Clustering Geometric methods, probabilities
Computer science(artificial intelligence) symbolic learning Pattern Recognition Description factorial Neural networks, genetic algorithms ...
Help for decision. Make relevant connections between data. Improve customer satisfaction Development of new products Can increase revenues while reducing costs
Data mining is totally dependent on the database which is being analyzed. Overconfidence: we should not follow blindly the result of a data mining analysis. It require qualified personnel: The data mining is a complex process that requires skilled staff .
20
Studying consumer behavior To maximize sales, many companies use data mining , to determine customer behavior and then aim their desires, so making more profits.
21
Data mining is used for other purposes than making profit, it is sometimes used to study social or medical parameters.
22
23