Methods and apparatus to partition data
View Patent ↗Methods and apparatus to partition data are discloses. An example method includes generating, with a processor, an indicator matrix for a set of panelist data based on a set of matrix criteria corresponding to the panelist data. The entries in the indicator matrix are determined based on a conditional probability of a panelist having one or more characteristics. An indicator is placed in a panelist vector of the matrix if the panelist has the one or more characteristics. A set of unique panelist vectors is determined and redundant panelist vectors that are not unique are removed to determine a minimum set of mutually exclusive partitions of the panelist data.
1. A method to reduce data storage requirements for panelist data stored in memory comprising:
generating, with a processor, an indicator matrix for a set of the panelist data stored in a memory by:
determining a first matrix criterion including a first and second panelist characteristic, the first matrix criterion being a probability that a panelist has the first characteristic given that the panelist has the second characteristic;
determining a second matrix criterion including a third and fourth panelist characteristic, the second matrix criterion being a probability that the panelist has the third characteristic given that the panelist has the fourth characteristic;
determining a first set of entries to be placed in a first element vector based on the first matrix criterion, the first set of entries including an entry for each respective panelist in the set of panelist data, each entry in the first set of entries to indicate whether each respective panelist has both the first panelist characteristic and the second panelist characteristic;
determining a second set of entries to be placed in a second element vector based on the first matrix criterion, the second set of entries including an entry for each respective panelist in the set of panelist data, each entry in the second set of entries to indicate whether each respective panelist has the second panelist characteristic;
determining a third set of entries to be placed in a third element vector based on the second matrix criterion, the third set of entries including an entry for each respective panelist in the set of panelist data, each entry in the third set of entries to indicate whether each respective panelist has both the third panelist characteristic and the fourth panelist characteristic;
determining a fourth set of entries to be placed in a fourth element vector based on the second matrix criterion, the fourth set of entries including an entry for each respective panelist in the set of panelist data, each entry in the fourth set of entries to indicate whether each respective panelist has the fourth panelist characteristic, the entries of the first, second, third, and fourth sets of entries corresponding to a same panelist and defining a panelist vector for the panelist, the panelist vectors and the element vectors defining the indicator matrix;
determining, with the processor, a set of unique panelist vectors, each unique panelist vector including a unique combination of the first, second, third, and fourth entries present in the indicator matrix; and
removing, with the processor, the set of panelist data and redundant panelist vectors that are not unique from the memory to determine a set of partitions of the panelist data to thereby reduce the data storage requirement for the panelist data.
2. The method of claim 1 , further including:
matching a subset of the panelist vectors with a matching one of the partitions to create a group of matching panelists;
determining a fifth characteristic for each of the matching panelists, the fifth characteristic not included in the element vectors; and
assigning a label of the one of the partitions to the panelists in the group.
3. The method of claim 2 , further including creating descriptions for the labels.
4. The method of claim 3 , wherein the descriptions include a respective matrix criterion included in the partition corresponding to the label.
5. The method of claim 2 , further including sorting the panelists based on the labels.
6. The method of claim 1 , further including receiving the first matrix criterion and the second matrix criterion based on a user input.
7. The method of claim 1 , wherein a number of the element vectors utilized in the indicator matrix corresponds to a number of matrix criteria to be analyzed.
8. The method of claim 1 , further including sorting the partitions based on labels assigned to the partitions.
9. A tangible computer readable storage medium comprising instructions that, when executed, cause a machine to at least:
generate, with a processor, an indicator matrix for a set of panelist data stored in a memory by:
determining a first matrix criterion including a first and second panelist characteristic, the first matrix criterion being a probability that a panelist has the first characteristic given that the panelist has the second characteristic;
determining a second matrix criterion including a third and fourth panelist characteristic, the second matrix criterion being a probability that the panelist has the third characteristic given that the panelist has the fourth characteristic;
determining a first set of entries to be placed in a first element vector based on the first matrix criterion, the first set of entries including an entry for each respective panelist in the set of panelist data, each entry in the first set of entries to indicate whether each respective panelist has both the first panelist characteristic and the second panelist characteristic;
determining a second set of entries to be placed in a second element vector based on the first matrix criterion, the second set of entries including an entry for each respective panelist in the set of panelist data, each entry in the second set of entries to indicate whether each respective panelist has the second panelist characteristic;
determining a third set of entries to be placed in a third element vector based on the second matrix criterion, the third set of entries including an entry for each respective panelist in the set of panelist data, each entry in the third set of entries to indicate whether each respective panelist has both the third panelist characteristic and the fourth panelist characteristic;
determining a fourth set of entries to be placed in a fourth element vector based on the second matrix criterion, the fourth set of entries including an entry for each respective panelist in the set of panelist data, each entry in the fourth set of entries to indicate whether each respective panelist has the fourth panelist characteristic, the entries of the first, second, third, and fourth sets of entries corresponding to a same panelist and defining a panelist vector for the panelist, the panelist vectors and the element vectors defining the indicator matrix;
determine, with the processor, a set of unique panelist vectors, each unique panelist vector including a unique combination of the first, second, third, and fourth entries present in the indicator matrix; and
remove, with the processor, the set of panelist data and redundant panelist vectors that are not unique from the memory to determine a set of partitions of the panelist data to thereby reduce the data storage requirement of the panelist data.
10. The storage medium of claim 9 , wherein the instructions further cause the machine to:
match a subset of the panelist vectors with a matching one of the partitions to create a group of matching panelists;
determine a fifth characteristic for each of the matching panelists, the fifth characteristic not included in the element vectors; and
assign a label of the one of the partitions to the panelists in the group.
11. The storage medium of claim 10 , wherein the instructions further cause the machine to create descriptions for the labels.
12. The storage medium of claim 11 , wherein the descriptions include a respective matrix criterion included in the partition corresponding to the label.
13. The storage medium of claim 12 , wherein the instructions further cause the machine to sort the panelists based on the labels.
14. The storage medium of claim 12 , wherein the instructions further cause the machine to sort the partitions based on labels assigned to the partitions.
15. The storage medium of claim 9 , wherein the instructions further cause the machine to receive the first matrix criterion and the second matrix criterion based on a user input.
16. The storage medium of claim 9 , wherein a number of the element vectors utilized in the indicator matrix corresponds to a number of matrix criteria to be analyzed.
17. A data partitioner for partitioning panelist data stored in memory comprising:
an indicator matrix generator to generate an indicator matrix for a set of panelist data stored in a memory by:
determining a first matrix criterion including a first and second panelist characteristic, the first matrix criterion being a probability that a panelist has the first characteristic given that the panelist has the second characteristic;
determining a second matrix criterion including a third and fourth panelist characteristic, the second matrix criterion being a probability that the panelist has the third characteristic given that the panelist has the fourth characteristic;
determining a first set of entries to be placed in a first element vector based on the first matrix criterion, the first set of entries including an entry for each respective panelist in the set of panelist data, each entry in the first set of entries to indicate whether each respective panelist has both the first panelist characteristic and the second panelist characteristic;
determining a second set of entries to be placed in a second element vector based on the first matrix criterion, the second set of entries including an entry for each respective panelist in the set of panelist data, each entry in the second set of entries to indicate whether each respective panelist has the second panelist characteristic;
determining a third set of entries to be placed in a third element vector based on the second matrix criterion, the third set of entries including an entry for each respective panelist in the set of panelist data, each entry in the third set of entries to indicate whether each respective panelist has both the third panelist characteristic and the fourth panelist characteristic;
determining a fourth set of entries to be placed in a fourth element vector based on the second matrix criterion, the fourth set of entries including an entry for each respective panelist in the set of panelist data, each entry in the fourth set of entries to indicate whether each respective panelist has the fourth panelist characteristic, the entries of the first, second, third, and fourth sets of entries corresponding to a same panelist and defining a panelist vector for the panelist, the panelist vectors and the element vectors defining the indicator matrix; and
a matrix reducer to determine a set of unique panelist vectors, each unique panelist vector including a unique combination of the first, second, third, and fourth entries present in the indicator matrix, and remove the set of the panelist data and redundant panelist vectors from the memory that are not unique to determine a set of partitions of the panelist data to thereby reduce the data storage requirement for the panelist data.
18. The data partitioner of claim 17 , further comprising:
the indicator matrix generator to determine a fifth characteristic for each of the matching panelists, the fifth characteristic not included in the element vectors;
a panelist sorter to match a subset of the panelist vectors with a matching one of the partitions to create a group of matching panelists; and
a panelist labeler to assign a label of the one of the partitions to the panelists in the group.
19. The apparatus of claim 17 , further including a data partitioner to receive the first matrix criterion and the second matrix criterion based on a user input.
20. The apparatus of claim 19 , wherein the panelist labeler is to create descriptions for the labels.