/**
* 1.按照过滤条件读取数据<br>
* 2.按照任务配置里设定的离散化方法，进行离散化<br>
* 3.按照任务配置里的时间粒度，对数据进行聚合
*
* @param filterCondition 过滤条件
* @param doAggregate 是否按时间对数据进行聚合
* @param doDiscretize 是否考虑离散化
* @return
*/
private DataItems readInput(String filterCondition, boolean doAggregate, boolean doDiscretize) {
String sqlStr =
"SELECT 事件发生时间," + task.getMiningObject() + " " + "FROM " + conn.DB_TABLE + " WHERE ";
if (task.getFilterCondition().length() > 0) sqlStr += task.getFilterCondition() + " AND ";
if (filterCondition != null && filterCondition.length() > 0)
sqlStr += filterCondition + " AND ";
sqlStr += "1=1 ORDER BY 事件发生时间 asc"; // 按时间先后顺序读取数据
conn.closeConn();
ResultSet rs = conn.sqlQuery(sqlStr);
if (rs == null) {
return null;
}
ResultSetMetaData meta = null;
int numRecords = 0;
try {
meta = rs.getMetaData();
int numCols = meta.getColumnCount();
data = new DataItems();
while (rs.next()) {
numRecords++;
StringBuilder sb = new StringBuilder();
for (int i = 2; i <= numCols; i++)
if (rs.getString(i) != null) sb.append(rs.getString(i).trim() + ",");
if (sb.length() > 0) {
Date d = parseTime(rs.getString(1).trim());
if (d != null) data.add1Data(d, sb.substring(0, sb.length() - 1));
else System.out.println("");
}
}
rs.close();
} catch (SQLException e) {
e.printStackTrace();
}
System.out.println("共" + numRecords + "条记录！");
System.out.println("读取完毕:" + data.getLength() + "条记录！");
boolean isNonDouble = !data.isAllDataIsDouble();
// 先进行时间粒度上的聚合
if (doAggregate)
data =
DataPretreatment.aggregateData(
data, task.getGranularity(), task.getAggregateMethod(), isNonDouble);
// 再进行离散化（只有数值型才能够离散化，否则应该会报错！）
if (doDiscretize)
data =
DataPretreatment.toDiscreteNumbers(
data,
task.getDiscreteMethod(),
task.getDiscreteDimension(),
task.getDiscreteEndNodes());
data.setGranularity(task.getGranularity()); // 设置数据的一些参数，如粒度
String endNodes = data.discreteNodes();
task.setDiscreteEndNodes(endNodes);
return data;
}
