In this article, we use Weka as our tool for data mining. In the first step, we retrieve the dataset from the UCI database. At the same time, we get the purpose of analysis. Then we classify the datasets and found class attributes. We classify the datasets into two attributes: latitude and longitude. The second step, we finish the data cleaning. The tools used for data cleansing are Microsoft Excel, Google Maps, and Weka. The next is aggregation and Skewed Data. Then the appropriate attributes of dataset are selected. The third step, it’s the experiment design. We choose three classifiers: Naive Bayes, J48 and IBk. The fourth step, we finally get the experimental results through the appropriate classifiers, and the results are summarized. The fifth step, to make ROC Curve. The sixth step, the analysis of final results. Of three aspects: classifier analysis, attribute analysis, noise analysis. In the seventh step, we get the final conclusion that IB1 is the most successful model for our dataset.
展开▼
