What's a data sample?
Data Sampling is a core technology in statistical analysis and machine learning. It refers to the process of selecting a subset (samples) from a large data set (general) in accordance with certain rules. By analysing the samples, we can extrapolate the overall characteristics without processing the full amount of data, thus significantly increasing the efficiency of calculations and reducing costs.
Why do data sampling have to be done?
- Increase processing speed: For millions or even tens of millions of lines of big data, direct analysis can lead to software collapse or very slow running. Samples provide quick previews.
- Reduce resource consumption: When expensive machine learning model training or A/B testing is conducted, the use of representative samples can significantly reduce server computing consumption.
- Research and audit• In quality testing or market research, it is not possible to examine each individual, and scientific samples ensure objectivity of results.
Sample algorithms supported by this tool
- Simple random sampling (Simple Random Sampling): Each individual is selected with the same probability and without prejudice.
- Percentage sample: Randomly extract data based on a given ratio (e.g. 10%) and apply to dynamic sampling when the number of the total lines is not determined.
- Stymmetric Sampling: The extraction is based on a fixed step (e.g., a line every 10 lines). If the data are sorted in time, the equidistance sampling can well cover the time dimension.
Privacy and security
Our online data sampling tool uses purely client-based technology. All your data are processed only in the browser memory and will never be uploaded to the back server. This ensures that when dealing with data sets containing personal privacy or commercial secrets, your information is 100% secure.