我已经通过k-means为终止条件定义了一个过滤器。 如果我运行我的应用程序,它总是只计算一次迭代。
我认为问题在于:
var positionWatchId;
function addWatch(){
positionWatchId = navigator.geolocation.watchPosition(onWatchPositionSuccess, onWatchPositionError, options);
}
function clearWatch(){
navigator.geolocation.clearWatch(positionWatchId);
}
function onWatchPositionError(err) {
window.plugins.toast.showLongBottom('App is waiting for location service! ', function (a) {
}, function (b) {
})
if (watchpositionErrorCount >= 9) {
clearWatch();
addWatch();
}
watchpositionErrorCount++;
}
addWatch();
或者过滤器功能:
DataSet<GeoTimeDataCenter> finalCentroids = loop.closeWith(newCentroids, newCentroids.join(loop).where("*").equalTo("*").filter(new MyFilter()));
最好的问候, 保罗
我的完整代码:
public static final class MyFilter implements FilterFunction<Tuple2<GeoTimeDataCenter, GeoTimeDataCenter>> {
private static final long serialVersionUID = 5868635346889117617L;
public boolean filter(Tuple2<GeoTimeDataCenter, GeoTimeDataCenter> tuple) throws Exception {
if(tuple.f0.equals(tuple.f1)) {
return true;
}
else {
return false;
}
}
}
答案 0 :(得分:4)
我认为问题是过滤功能(以你未发布的代码为模)。 Flink的终止标准以下列方式工作:如果提供的终止DataSet
为空,则满足终止标准。否则,如果未超过最大迭代次数,则启动下一次迭代。
Flink的filter
函数仅保留FilterFunction
返回true
的元素。因此,使用MyFilter
实现,您只能保持迭代前后的质心相同。这意味着如果所有质心都已更改,您将获得空DataSet
,因此迭代终止。这显然与实际终止标准相反。终止标准应该是:只要存在已经改变的质心,就用k-means继续。
您可以使用coGroup
函数执行此操作,如果前一个质心DataSet
没有匹配的质心,则会发出元素。这类似于左外连接,只是您放弃非空匹配。
public static void main(String[] args) throws Exception {
// set up the execution environment
final ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();
DataSet<Element> oldDS = env.fromElements(new Element(1, "test"), new Element(2, "test"), new Element(3, "foobar"));
DataSet<Element> newDS = env.fromElements(new Element(1, "test"), new Element(3, "foobar"), new Element(4, "test"));
DataSet<Element> filtered = newDS.coGroup(oldDS).where("*").equalTo("*").with(new FilterCoGroup());
filtered.print();
}
public static class FilterCoGroup implements CoGroupFunction<Element, Element, Element> {
@Override
public void coGroup(
Iterable<Element> newElements,
Iterable<Element> oldElements,
Collector<Element> collector) throws Exception {
List<Element> persistedElements = new ArrayList<Element>();
for(Element element: oldElements) {
persistedElements.add(element);
}
for(Element newElement: newElements) {
boolean contained = false;
for(Element oldElement: persistedElements) {
if(newElement.equals(oldElement)){
contained = true;
}
}
if(!contained) {
collector.collect(newElement);
}
}
}
}
public static class Element implements Key {
private int id;
private String name;
public Element(int id, String name) {
this.id = id;
this.name = name;
}
public Element() {
this(-1, "");
}
@Override
public int hashCode() {
return 31 + 7 * name.hashCode() + 11 * id;
}
@Override
public boolean equals(Object obj) {
if(obj instanceof Element) {
Element element = (Element) obj;
return id == element.id && name.equals(element.name);
} else {
return false;
}
}
@Override
public int compareTo(Object o) {
if(o instanceof Element) {
Element element = (Element) o;
if(id == element.id) {
return name.compareTo(element.name);
} else {
return id - element.id;
}
} else {
throw new RuntimeException("Comparing incompatible types.");
}
}
@Override
public void write(DataOutputView dataOutputView) throws IOException {
dataOutputView.writeInt(id);
dataOutputView.writeUTF(name);
}
@Override
public void read(DataInputView dataInputView) throws IOException {
id = dataInputView.readInt();
name = dataInputView.readUTF();
}
@Override
public String toString() {
return "(" + id + "; " + name + ")";
}
}