I have trained TokenNameFinder of OpenNLP which outputs .bin file. Now I need to list features with their importance.
I read code of TokenNameFinder and NameFinderME but could not find a way to print features. Is their any way to list all features of model along with their importance ?
Finally I figured out a way to list features. Function getDataStructures() of AbstractModel class returns a array of Object instances. Second element of this array is a Map<String, Integer> whose keys are combination of features and their values. Following is the code snippet for accessing features and their values:
AbstractModel maxModel = model.getArtifact("nameFinder.model");
Object[] obj = maxModel.getDataStructures();
if(obj!=null) {
Map<String, Integer> pmap = (HashMap<String, Integer>) obj[1];
Set<String> keySet = pmap.keySet();
for(String key: keySet) {
System.out.println(key +" **** "+ pmap.get(key));
}
} else {
System.out.println("obj is null." );
}
Related
I have the below class:
class A{
String property1;
String property2;
Double property3;
Double property4;
}
So the property1 and property2 is the key.
class Key{
String property1;
String property2;
}
I already have a list of A like below:
List<A> list=new ArrayList<>();
I want to group by using the key and add to another list of A in order to avoid having multiple items with same key in the list:
Function<A, Key> keyFunction= r-> Key.valueOf(r.getProperty1(), r.getProperty2());
But then while doing group by I have to take a sum of property3 and average of property4.
I need an efficient way to do it.
Note: I have skipped the methods of the given classes.
Collecting to a Map is unavoidable since you want to group things. A brute-force way to do that would be :
yourListOfA
.stream()
.collect(Collectors.groupingBy(
x -> new Key(x.getProperty1(), x.getProperty2()),
Collectors.collectingAndThen(Collectors.toList(),
list -> {
double first = list.stream().mapToDouble(A::getProperty3).sum();
// or any other default
double second = list.stream().mapToDouble(A::getProperty4).average().orElse(0D);
A a = list.get(0);
return new A(a.getProperty1(), a.getProperty2(), first, second);
})))
.values();
This could be slightly improved for example in the Collectors.collectingAndThen to only iterate the List once, for that a custom collector would be required. Not that complicated to write one...
Try like this:
Map<A,List<A>> map = aList
.stream()
.collect(Collectors
.groupingBy(item->new A(item.property1,item.property2)));
List<A> result= map.entrySet().stream()
.map(list->new A(list.getValue().get(0).property1,list.getValue().get(0).property1)
.avgProperty4(list.getValue())
.sumProperty3(list.getValue()))
.collect(Collectors.toList());
and create avgProperty4 and sumProperty3 methods like to this
public A sumProperty3(List<A> a){
this.property3 = a.stream().mapToDouble(A::getProperty3).sum();
return this;
}
public A avgProperty4(List<A> a){
this.property4 = a.stream().mapToDouble(A::getProperty4).average().getAsDouble();
return this;
}
result = aList.stream().collect(Collectors
.groupingBy(item -> new A(item.property1, item.property2),
Collectors.collectingAndThen(Collectors.toList(), list ->
new A(list.get(0).property1, list.get(0).property1)
.avgProperty4(list).sumProperty3(list))
)
);
I've got some working, inelegant code here:
The custom object is:
public class Person {
private int id;
public getId() { return this.id }
}
And I have a Class containing a Set<Person> allPersons containing all available subjects. I want to extract a new Set<Person> based upon one or more ID's of my choosing. I've written something which works using a nested enhanced for loop, but it strikes me as inefficient and will make a lot of unnecessary comparisons. I am getting used to working with Java 8, but can't quite figure out how to compare the Set against an Array. Here is my working, but verbose code:
public class MyProgram {
private Set<Person> allPersons; // contains 100 people with Ids 1-100
public Set<Person> getPersonById(int[] ids) {
Set<Person> personSet = new HashSet<>() //or any type of set
for (int i : ids) {
for (Person p : allPersons) {
if (p.getId() == i) {
personSet.add(p);
}
}
}
return personSet;
}
}
And to get my result, I'd call something along the lines of:
Set<Person> resultSet = getPersonById(int[] intArray = {2, 56, 66});
//resultSet would then contain 3 people with the corresponding ID
My question is how would i convert the getPersonById method to something using which streams allPersons and finds the ID match of any one of the ints in its parameter array? I thought of some filter operation, but since the parameter is an array, I can't get it to take just the one I want only.
The working answer to this is:
return allPersons.stream()
.filter(p -> (Arrays.stream(ids).anyMatch(i -> i == p.getId())) )
.collect(Collectors.toSet());
However, using the bottom half of #Flown's suggestion and if the program was designed to have a Map - it would also work (and work much more efficiently)
As you said, you can introduce a Stream::filter step using a Stream::anyMatch operation.
public Set<Person> getPersonById(int[] ids) {
Objects.requireNonNull(ids);
if (ids.length == 0) {
return Collections.emptySet();
}
return allPersons.stream()
.filter(p -> IntStream.of(ids).anyMatch(i -> i == p.getId()))
.collect(Collectors.toSet());
}
If the method is called more often, then it would be a good idea to map each Person to its id having a Map<Integer, Person>. The advantage is, that the lookup is much faster than iterating over the whole set of Person.Then your algorithm may look like this:
private Map<Integer, Person> idMapping;
public Set<Person> getPersonById(int[] ids) {
Objects.requireNonNull(ids);
return IntStream.of(ids)
.filter(idMapping::containsKey)
.mapToObj(idMapping::get)
.collect(Collectors.toSet());
}
I have map of a map of strings. This map is a parsing of a json object and represents the criteria entered by the user to filter a list in the UI.
In the rest service I want to populate an object with data comes from this map. Unfortunately I cannot change queryModel Object. Query Model object has a list of filters. Each filter has a list of fields and a list of operations to be applied to the field. My goal is to convert the following code with java 8 stream.
for(Map.Entry<String,Map<String,String>> entry: filters.entrySet()) {
Filter filter = new Filter();
filter.setFields(new ArrayList<String>());
filter.getFields().add(entry.getKey());
filter.setValues(new ArrayList<String>());
filter.setOperators(new ArrayList<String>());
if (entry.getValue() != null) {
for(String key : entry.getValue().keySet()) {
if(key.equals("value")) {
filter.getValues().add(entry.getValue().get(key));
}
else if(key.equals("matchMode")){
filter.getOperators().add(entry.getValue().get(key));
}
}
queryModel.getFilters().add(filter);
}
As you can see I first set the name of the field in the fields list and then for that field I loop in the values to get the value entered and the match mode. In a functional I don't know ho to save the field of the outer loop to set it in the filter object created in the inner loop.
That was my attempt
public static Filter getFilter(Map.Entry<String,String> entry) {
Filter filter = new Filter();
filter.setFields(new ArrayList<String>());
filter.getFields().add(entry.getKey());
filter.setValues(new ArrayList<String>());
filter.setOperators(new ArrayList<String>());
if(entry.getKey().equals("value")) {
filter.getValues().add(entry.getValue());
}
else if(entry.getKey().equals("matchMode")){
filter.getOperators().add(entry.getValue());
}
return filter;
}
List<Filter> filterList = filters.entrySet().stream()
.filter( stringMapEntry -> stringMapEntry.getValue() != null)
.flatMap( entry -> entry.getValue().entrySet().stream())
.map (innerEntry-> QueryModelAdapter.getFilter(innerEntry))
.collect (Collectors.toList());
queryModel.setFilters (filterList);
I need in QueryModelAdapter.getFilter the entry of the flat map. How can I do that?
Before I say anything, be polite when asking questions. Nobody gets paid for answering questions here. All are doing it for their pleasure.
So, be nice to them at least with your words.
Alright, I think your question is more suitable for CodeReview than StackOverflow.
One thing to note, You can't rewrite your legacy java projects to have every single line with lambdas and streams.
Sometimes, it's better the old fashioned way than the new features.
You don't need to iterate a Map to retrieve its matching value. You can remove that Inner-loop.
Let's take your current class (whatever the class you copied the code from) named it as RespectOthers.java
private static Filter getEmptyFilter(){
Filter filter = new Filter();
filter.setFields(new ArrayList<String>());
filter.setValues(new ArrayList<String>());
filter.setOperators(new ArrayList<String>());
return filter;
}
private static Filter setKeyAndValues(Filter inputFilterObj, Map.Entry<String,Map<String,String>> entry, QueryModel queryModel){
inputFilterObj.setFields(new ArrayList<String>());
inputFilterObj.getFields().add(entry.getKey());
if (entry.getValue() != null) {
inputFilterObj.getValues().add(entry.getValue().get("value"));
inputFilterObj.getOperators().add(entry.getValue().get("matchMode"));
queryModel.getFilters().add(inputFilterObj);
}
return inputFilterObj;
}
List<Filter> finalOutput = filters.entrySet().stream()
.map(e -> RespectOthers.setKeyAndValues(RespectOthers.getEmptyFilter(), e, myQueryModel))
.collect(Collectors.toList());
I am trying to make a List of all of the books in one Collection that are not present in another. My problem is that I need to compare based on book ID, so I can't just test to see whether a book in the first is contained in the second, I have to determine whether any book in the second collection has the same ID as a book in the first.
I have the below code to compare two collections of books and filter the first collection:
List<Book> parentBooks = listOfBooks1.stream().filter(book->
!listOfBooks2.contains(book)).collect(Collectors.toList());
The code doesn't work correctly because I am comparing the objects themselves. I need to compare the objects based on the bookId instead of the whole book object. How should I change the code so it can do the comparison based on the bookId (book.getId())?
List<Book> books1 = ...;
List<Book> books2 = ...;
Set<Integer> ids = books2.stream()
.map(Book::getId)
.collect(Collectors.toSet());
List<Book> parentBooks = books1.stream()
.filter(book -> !ids.contains(book.getId()))
.collect(Collectors.toList());
The problem is complex, but it boils down to one thing, knows your data. Is it immutables, entities with an id, duplicate entries etc?
The code below works for immutables with only values (and with possible duplicates).
It first tries to remove all entries in the before list (from the copied after-list).
What is left will be the added elements. The ones from the before-list that can be removed from the after-list are the unchanged ones.
The rest are the removed ones
public class ListDiffer<T> {
private List<T> addedList = new ArrayList<>();
private List<T> unchangedList = new ArrayList<>();
private List<T> removedList = new ArrayList<>();
public ListDiffer(List<T> beforeList, List<T> afterList) {
addedList.addAll(afterList); // Will contain only new elements when all elements in the Before-list are removed.
beforeList.forEach(e -> {
boolean b = addedList.remove(e) ? unchangedList.add(e) : removedList.add(e);
});
}
public List<T> getAddedList() {
return addedList;
}
public List<T> getUnchangedList() {
return unchangedList;
}
public List<T> getRemovedList() {
return removedList;
}
}
getSimpleJdbcTemplate().query(sql, getMapper()); returns List, but I need a Map where key will be store data of one of the field of object. For example, I have object named "Currency" which has fields: id, code, name, etc. Code above will return List object, but I want to get currency by id from Map. Now, I wrote the following code:
#Override
public Map<Integer, Currency> listCurrencies() {
String sql = "select cur_id, cur_code, cur_name ... from currencies";
List<Currency> currencies = getSimpleJdbcTemplate().query(sql, getMapper());
Map<Integer, Currency> map = new HashMap<Integer, Currency>(currencies.size());
for (Currency currency : currencies) {
map.put(currency.getId(), currency);
}
return map;
}
Are there any way to do same but without creating List object and looping inside it?
You have ResultSetExtractor for extracting values from the ResultSet. So in your case you can write a custom ResultSetExtractor which will return you the Map object.