SliceQuery Class

NReco.PivotData Class Library Documentation

Represents a query operation over specified IPivotData instance.

Inheritance Hierarchy

System.Object
NReco.PivotData.SliceQuery

Namespace: NReco.PivotData
Assembly: NReco.PivotData (in NReco.PivotData.dll) Version: 1.4.1

Syntax

C++

Copy

public class SliceQuery

Public Class SliceQuery

public ref class SliceQuery

type SliceQuery =  class end

The SliceQuery type exposes the following members.

Constructors

	Name	Description
	SliceQuery	Initializes new slicing query to specified PivotData instance.

Top

Methods

	Name	Description
	Dimension(String)	Define dimension to select in result of this query.
	Dimension(String, Func<Object[], Object>)	Define dimension constructed from keys of existing dimensions.
	Equals	(Inherited from Object.)
	Execute()	Execute the query and return operation result as new PivotData instance.
	Execute(Boolean)	Execute the query and return operation result as new PivotData instance.
	Finalize	(Inherited from Object.)
	GetHashCode	(Inherited from Object.)
	GetType	(Inherited from Object.)
	Measure(Int32)	Define measure aggregator at specified index to select in result of this query.
	Measure(IAggregatorFactory, Func<IAggregator, IAggregator>)	Define new measure aggregator calculated from existing cube measure(s).
	Measure(IAggregatorFactory, Func<KeyValuePair<Object[], IAggregator>, IAggregator>)	Define new measure aggregator calculated from existing cube measure(s).
	Measure(String, Func<IAggregator[], Object>,Int32[])	Define formula measure.
	MemberwiseClone	(Inherited from Object.)
	ToString	(Inherited from Object.)
	Where(Func<KeyValuePair<Object[], IAggregator>, Boolean>)	Filters data points based on a predicate.
	Where(String, Func<Object, Boolean>)	Filters dimension keys based on a predicate.
	Where(String,Object[])	Filters dimension keys by explicit list of values.

Top

Remarks

SliceQuery can be used for querying data cube (slice, dice, roll-up).

Examples

Lets assume that we have some cube that represents sales data with the following dimensions: "year", "month", "day", "country", "product" and 2 measures collected with CountAggregatorFactory (=count of orders) and SumAggregatorFactory (= sum of sales amount):

Copy

var salesCube = new PivotData(
                new [] {"year","month","day","country","product"},
                new CompositeAggregatorFactory(
                    new CountAggregatorFactory(),
                    new SumAggregatorFactory("amount")
                ), true );

The following query illustrates how to reduce number of dimensions, filter by specific dimension values and get resulting PivotData with single measure:

Copy

var q = new SliceQuery(salesCube)
                .Dimension("year")
                .Dimension("country")
                .Dimension("product")
                .Where("country", new[]{"USA","Canada"})
                .Measure(1);
            var salesAmountForUsaAndCanada = q.Execute(true);

Resulting data cube will contain only specified dimensions, "country" dimension will contain only "USA" and "Canada" keys and only one measure (index=1 refers to SumAggregatorFactory("amount")).

Examples

SliceQuery can be used in the more complex filtering cases; lets calculate derived dimension and skip all days with <10 orders:

Copy

var q = new SliceQuery(salesCube)
                .Dimension("year")
                .Dimension("quarter", (dimKeys) => {  
                    var month = Convert.ToInt32( dimKeys[1] ); // "month" dimension index
                    return (int)Math.Ceiling((float)(month)/3);
                }).Where( (dataPoint) => {
                    var compositeAggr = dataPoint.Value.AsComposite(); // CompositeAggregator is used if cube has >1 measure
                    var countOrders = Convert.ToInt32( compositeAggr.Aggregators[0].Value );
                    return countOrders>=10; // include data point by custom condition
                });
            var bigSalesByYearAndQuarter = q.Execute(true);

Reference

NReco.PivotData Namespace