Estudo comparativo entre algoritmos de regras de associação de forma normal e incremental de dados

AUTOR(ES)
DATA DE PUBLICAÇÃO

2008

RESUMO

Now a days the biggest challenge to the Information Technology is to catch a glimpse of the data mining exploration mechanisms to present a detailed analysis with the objective to detect new discoveries in business or market tendencies. Aggregated to this necessity has surged several technicians proposing to realize this job, called Data Mining Technician. Among those, the data mining detach the Rules Association, defined as descriptive standards representing the probability of a item set to appear in a transaction. In this dissertation were explored two association rules algorithms, the APriori, applied to rules extraction in conventional way and the ZigZag algorithm, utilized to extract rules association in incremental way, which the main objective is to effectuate a comparative research between these algorithms as well as its utilization. Both was applied in two data base in health insurance company to demonstrate a comparison of the results, measuring time, amount and quality of the extracted rules from both data base. In all algorithms extractions the ZigZag has showed it self effective in its performance, in comparison to the APriori applications. However the results analysis is not a trivial duty, tending to subjective results. To this, was done many jobs of interpretation of the results as well applied measurements of the rules evaluations with the purpose to obtain the best quality within the extracted rules and its respective algorithms and data bases.

ASSUNTO(S)

exploração de dados data mining computer algorithms algoritmos de computador ciencia da computacao informática - dissertações

ACESSO AO ARTIGO

Documentos Relacionados