Reducing a number of units in a data population based on predefined constraints
A method includes receiving a data population and a plurality of constraints. The data population includes a plurality of units. Each unit includes a plurality of respective values. Each value corresponds to a respective parameter. Each constraint corresponds to a respective parameter. A holistic feature of the data population is determined. A volume is defined in the parameter space based on the plurality of constraints. For each unit of a current data population, a first vector representing a respective unit is subtracted from a second vector representing a current holistic feature to determine a third vector. For each third vector, a distance between a point defined by a respective third vector and the volume is determined. A minimum distance is determined from determined distances. The current data population is transformed into an updated data population by removing a unit that corresponds to the minimum distance from the current data population.
1. A system comprising:
a memory configured to store:
a data population, wherein:
the data population comprises a plurality of units;
each unit comprises a plurality of respective values; and
each value corresponds to a respective parameter; and
a plurality of constraints, wherein each constraint corresponds to a respective parameter; and
a processor communicatively coupled to the memory, wherein the processor is configured to:
receive the data population;
receive the plurality of constraints;
represent the units of the data population by first vectors in a parameter space;
determine a holistic feature of the data population;
represent the holistic feature by a second vector in the parameter space, wherein the second vector is a linear combination of the first vectors;
define a volume in the parameter space based on the plurality of constraints, the volume having a multi-dimensional shape;
(a) for each unit of a current data population, subtract a vector representing a respective unit from a vector representing a current holistic feature to determine a third vector, wherein the vector representing the current holistic feature is a linear combination of vectors that represent units of the current data population;
(b) for each third vector, determine a distance between a point defined by a respective third vector and the volume;
(c) determine a minimum distance from determined distances;
(d) transform the current data population into an updated data population by removing a unit from the current data population that corresponds to the minimum distance from the current data population;
(e) determine an updated holistic feature of the updated data population;
repeat steps (a)-(e) until the updated holistic feature is within the volume; and
determine a final data population, the final data population being a subset of the received data population that comprises a maximum number of units of the data population that have the updated holistic feature within the volume.
2. The system of claim 1 , wherein a vector representing the updated holistic feature is a linear combination of vectors that represent units of the updated data population.
3. The system of claim 1 , wherein the holistic feature is determined based on all of the units of the data population.
4. The system of claim 1 , wherein each parameter comprises a financial parameter, a demographic parameter, or a computer parameter.
5. The system of claim 1 , wherein each constraint defines a minimum value and a maximum value for a respective parameter.
6. A method comprising:
receiving a data population, wherein:
the data population comprises a plurality of units;
each unit comprises a plurality of respective values; and
each value corresponds to a respective parameter;
receiving a plurality of constraints, wherein each constraint corresponds to a respective parameter;
representing the units of the data population by first vectors in a parameter space;
determining a holistic feature of the data population;
representing the holistic feature by a second vector in the parameter space, wherein the second vector is a linear combination of the first vectors;
defining a volume in the parameter space based on the plurality of constraints, the volume having a multi-dimensional shape;
(a) for each unit of a current data population, subtracting a vector representing a respective unit from a vector representing a current holistic feature to determine a third vector, wherein the vector representing the current holistic feature is a linear combination of vectors that represent units of the current data population;
(b) for each third vector, determining a distance between a point defined by a respective third vector and the volume;
(c) determining a minimum distance from determined distances;
(d) transforming the current data population into an updated data population by removing a unit from the current data population that corresponds to the minimum distance from the current data population;
(e) determining an updated holistic feature of the updated data population;
repeating steps (a)-(e) until the updated holistic feature is within the volume; and
determining a final data population, the final data population being a subset of the received data population that comprises a maximum number of units of the data population that have the updated holistic feature within the volume.
7. The method of claim 6 , wherein a vector representing the updated holistic feature is a linear combination of vectors that represent units of the updated data population.
8. The method of claim 6 , wherein the holistic feature is determined based on all of the units of the data population.
9. The method of claim 6 , wherein each parameter comprises a financial parameter, a demographic parameter, or a computer parameter.
10. The method of claim 6 , wherein each constraint defines a minimum value and a maximum value for a respective parameter.
11. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to:
receive a data population, wherein:
the data population comprises a plurality of units;
each unit comprises a plurality of respective values; and
each value corresponds to a respective parameter;
receive a plurality of constraints, wherein each constraint corresponds to a respective parameter;
represent the units of the data population by first vectors in a parameter space;
determine a holistic feature of the data population;
represent the holistic feature by a second vector in the parameter space, wherein the second vector is a linear combination of the first vectors;
define a volume in the parameter space based on the plurality of constraints, the volume having a multi-dimensional shape;
(a) for each unit of a current data population, subtract a vector representing a respective unit from a vector representing a current holistic feature to determine a third vector, wherein the vector representing the current holistic feature is a linear combination of vectors that represent units of the current data population;
(b) for each third vector, determine a distance between a point defined by a respective third vector and the volume;
(c) determine a minimum distance from determined distances;
(d) transform the current data population into an updated data population by removing a unit from the current data population that corresponds to the minimum distance from the current data population;
(e) determine an updated holistic feature of the updated data population;
repeat steps (a)-(e) until the updated holistic feature is within the volume; and
determine a final data population, the final data population being a subset of the received data population that comprises a maximum number of units of the data population that have the updated holistic feature within the volume.
12. The non-transitory computer-readable medium of claim 11 , wherein a vector representing the updated holistic feature is a linear combination of vectors that represent units of the updated data population.
13. The non-transitory computer-readable medium of claim 11 , wherein each parameter comprises a financial parameter, a demographic parameter, or a computer parameter.
14. The non-transitory computer-readable medium of claim 11 , wherein each constraint defines a minimum value and a maximum value for a respective parameter.