Data refining model based on oil refining process

As data volume rapidly increases during recent years, traditional data mining methods meet some problems. Considering the composition of oil is similar to that of data, we propose data refining based on oil refining to make the data clean and minable. There are physical changes and chemical changes in oil refining. We define data atmospheric-vacuum distillation as the physical changes and data catalytic cracking as the chemical changes. Data atmospheric-vacuum distillations just separate the original data into data fractions. And data catalytic cracking continues clustering the data and changes the data elements in the fraction. After data refining, the important data will be grouped into parts of the final clusters, and further mining can be adopted in these clusters. Finally, we use Shanghai dynamic 101 radio data to validate the effectiveness of data refining.