PR000005 practice question 23 of 25

When is sampling for Profiling a good idea? Select the best answer

Full set: 140 questions
Answer options
A:

A. When the data set contains sensitive information like credit card numbers.

B:

B. When the probability of data quality issues is small.

C:

C. When the client provides assurances the quality of the data is good.

D:

D. When the volume of data is large and there are storage and/or time constraints.

Correct answer
Correct Answer:
D. D. When the volume of data is large and there are storage and/or time constraints.
Explanation

Sampling is a performance optimization technique used in data profiling. When dealing with very large datasets (e.g., billions of rows), running a profile on the entire dataset can be extremely time-consuming and resource-intensive, potentially exceeding available processing windows or storage capacity for the results. By profiling a statistically significant sample, an organization can gain a reasonably accurate understanding of the data's structure, quality, and patterns in a fraction of the time and with fewer computational resources. This approach balances the need for data insight against practical constraints of time and infrastructure.

Why Incorrect

A. The presence of sensitive data is a security concern addressed by data masking or tokenization, not a primary reason to use sampling for profiling.
B. If the probability of data quality issues is small, sampling might miss these rare but potentially critical anomalies, making a full profile more appropriate.
C. Client assurances should be verified through comprehensive profiling; relying on them is not a valid technical reason to limit the scope of the analysis by sampling.

References

1. Informatica 10.5, Data Explorer Guide, Chapter 2: Profiling, Section: "Profile Sampling", Page 23. The documentation states, "You can run a profile on a sample of the source data to reduce the time that it takes to run the profile. You might want to run a profile on a sample of source data if the data object contains a very large number of rows."

2. Informatica Intelligent Cloud Services, Data Profiling Documentation, Section: "Sampling options". The guide explains, "To improve performance when you profile data, you can configure sampling options... Use sampling when you profile large data sets."

3. Abedjan, Z., Golab, L., & Naumann, F. (2015). Profiling Relational Data: A Survey. The VLDB Journal, 24(4), 557–581. Section 6, "Scalability," discusses challenges with large data volumes: "For very large databases, it may be infeasible to scan the entire data... In such cases, it is common to resort to data sampling." (https://doi.org/10.1007/s00778-015-0384-y)

Scroll to Top