Automated adaptive data analysis using dynamic data quality assessment
Embodiments provide for automated dynamic data quality assessment. A data quality job including a new data sample is received. If the new data sample is determined to be added to a reservoir of data samples, a quality verification request is sent to an oracle. A new data sample quality estimate is received from the oracle. The new data sample and estimate can be added to the reservoir. A judgment associated with a new data sample is received from a predictive model. Whether to send a new data sample quality verification request to an oracle is determined. If a new data sample quality estimate is received from the oracle, whether to add the new data sample and the judgment to the reservoir is determined.
1. An apparatus for automated adaptive data analysis, the apparatus comprising at least one processor and at least one memory storing instructions that, with the at least one processor, cause the apparatus to:
generate, using a predictive model, a judgment associated with a new data sample received in a data stream, wherein the judgment is generated based on a feature vector associated with the new data sample, and wherein the feature vector represents an optimal view of the new data sample; and
update reservoir summary statistics associated with a reservoir of data samples in response to determining whether to add the new data sample and the judgment to the reservoir of data samples, wherein the determining is based on one of whether the new data sample statistically belongs in the data reservoir or whether the judgment is associated with a high confidence value.
2. The apparatus of claim 1 , wherein the determining is in response to receiving a data quality estimate from an oracle.
3. The apparatus of claim 2 , wherein the at least one memory storing instructions that, with the at least one processor, cause the apparatus to:
send a quality verification request comprising the new data sample to the oracle.
4. The apparatus of claim 1 , wherein the reservoir of data samples is identified based at least in part on a data type of the new data sample.
5. The apparatus of claim 4 , wherein the predictive model is adapted using a set of training data samples having the data type.
6. The apparatus of claim 5 , wherein the at least one memory storing instructions, with the at least one processor, further cause the apparatus to:
update the set of training data samples in response to comparing the reservoir summary statistics to training data summary statistics derived from the set of training data samples having the data type and determining to update the set of training data samples.
7. The apparatus of claim 1 , wherein the at least one memory storing instructions that, with the at least one processor, cause the apparatus to:
generate, using the predictive model, a confidence value associated with the judgment.
8. The apparatus of claim 1 , wherein the at least one memory storing instructions that, with the at least one processor, cause the apparatus to:
replace the judgment with a second judgment associated with the new data sample received from an oracle when the judgment and second judgment do not match.
9. A computer readable storage medium comprising a non-transitory storage medium for storing instructions that, when executed by at least one processor, cause the processor to:
generate, using a predictive model, a judgment associated with a new data sample received in a data stream, wherein the judgment is generated based on a feature vector associated with the new data sample, and wherein the feature vector represents an optimal view of the new data sample; and
update reservoir summary statistics associated with a reservoir of data samples in response to determining whether to add the new data sample and the judgment to the reservoir of data samples, wherein the determining is based on one of whether the new data sample statistically belongs in the data reservoir or whether the judgment is associated with a high confidence value.
10. The computer readable storage medium of claim 9 , wherein the determining is in response to receiving a data quality estimate from an oracle.
11. The computer readable storage medium of claim 10 , wherein the at least one memory storing instructions that, with the at least one processor, cause the apparatus to:
send a quality verification request comprising the new data sample to the oracle.
12. The computer readable storage medium of claim 9 , wherein the reservoir of data samples is identified based at least in part on a data type of the new data sample.
13. The computer readable storage medium of claim 12 , wherein the predictive model is adapted using a set of training data samples having the data type.
14. The computer readable storage medium of claim 13 , comprising a non-transitory storage medium for storing instructions that, when executed by at least one processor, further cause the processor to:
update the set of training data samples in response to comparing the reservoir summary statistics to training data summary statistics derived from the set of training data samples having the data type and determining to update the set of training data samples.
15. The computer readable storage medium of claim 9 , comprising a non-transitory storage medium for storing instructions that, when executed by at least one processor, further cause the processor to:
generate, using the predictive model, a confidence value associated with the judgment.
16. The computer readable storage medium of claim 9 , comprising a non-transitory storage medium for storing instructions that, when executed by at least one processor, further cause the processor to:
replace the judgment with a second judgment associated with the new data sample received from an oracle when the judgment and second judgment do not match.
17. A computer implemented method, comprising:
generating, using at least one processor and a predictive model, a judgment associated with a new data sample received in a data stream, wherein the judgment is generated based on a feature vector associated with the new data sample, and wherein the feature vector represents an optimal view of the new data sample; and
updating, using the at least one processor, reservoir summary statistics associated with a reservoir of data samples in response to determining whether to add the new data sample and the judgment to the reservoir of data samples, wherein the determining is based on one of whether the new data sample statistically belongs in the data reservoir or whether the judgment is associated with a high confidence value.
18. The method of claim 17 , wherein the determining is in response to receiving a data quality estimate from an oracle.
19. The method of claim 17 , wherein the reservoir of data samples is identified based at least in part on a data type of the new data sample.
20. The method of claim 19 , wherein the predictive model is adapted using a set of training data samples having the data type.