In general, this would not typically be a concern of a data mining problem since the methods and software are intended for large data sets. Sampling can be helpful when resources or limited or when the distributions in the data are not reflective of the population, but the goal is to have your training/validation/test sample be as representative as possible.
If you are sampling, however, one approach would be to cluster the first data set, score the second data set using the cluster solution obtained on the first, and then sample proportionally from the second data set based on the distribution of clusters in the first data set. You might consider stratifying on certain grouping variables (e.g. gender, location, etc...) to make the distribution as balanced as possible.
I hope this helps!
Doug
In general, this would not typically be a concern of a data mining problem since the methods and software are intended for large data sets. Sampling can be helpful when resources or limited or when the distributions in the data are not reflective of the population, but the goal is to have your training/validation/test sample be as representative as possible.
If you are sampling, however, one approach would be to cluster the first data set, score the second data set using the cluster solution obtained on the first, and then sample proportionally from the second data set based on the distribution of clusters in the first data set. You might consider stratifying on certain grouping variables (e.g. gender, location, etc...) to make the distribution as balanced as possible.
I hope this helps!
Doug
Join us for SAS Innovate April 16-19 at the Aria in Las Vegas. Bring the team and save big with our group pricing for a limited time only.
Pre-conference courses and tutorials are filling up fast and are always a sellout. Register today to reserve your seat.
Use this tutorial as a handy guide to weigh the pros and cons of these commonly used machine learning algorithms.
Find more tutorials on the SAS Users YouTube channel.