Fairgen's pioneer data augmentation solution can generate near-real synthetic results into niche segments. As long as based on a model trained on your general audience survey.
Technical validation is a key part of every workflow in Fairgen.
Can anyone sign up?
Yes. Currently, any organization can apply online by scheduling an introductory call with our team. Be among the first to deliver unprecedented granularity using reliable and explainable AI technology,
What is the impact of Fairboost on Insights teams?
Insights teams can make better niche go-to-market decisions by having fast and affordable access to data that was before too hard, or impossible, to get. It provides insides teams with the ability to zoom in on sub-segments and gain more trust when reading these groups. Researchers can immediately expand coverage across niche markets, while gaining fast access do data and without the need of costly traditional data collection boosts. Access to once "impossible-to-reach" segments is an added value: by extrapolating results while maintaining high confidence levels, niche segments once considered too small, or too hard-to-reach for fieldwork boosts, can now be analyzed.
How is the technical validation process for Fairboost?
In short, you will run several parallel testing processes that will compare the efficiency of synthetic data versus real data.
The primary method used by our clients is "Parallel testing", using as a base existing datasets. By the end of it, our predictive technology's accuracy and coherence both on aggregated and granular results can be determined. The process is simple: you will provide a fraction of an existing sample from a given segment for Fairgen to be trained on, so we can mimic a situation where you have scarce data. Let's say 10% of it. The remaining of it will be used as the ground truth and Fairgen will generate a synthetic dataset that was trained on only 10% of it. We will then compare its reliability against samples of the "ground truth". We guarantee a reliable and coherent augmentation that will perform significantly better than what our models were trained on.
Can Fairgen handle any type of data?
Yes. Fairgen is equipped to handle any quantitative (close-ended) data points that can be exported in a columnar format (single and multiple choice questions, rankings, NPS, continuous open-ended, ratings, and more). Fairgen automatically recognizes field types and conditional relationships between questions.
Can Fairgen augment qualitative questions?
No. Fairgen’s can augment only quantitative fields.
How is it different than reweighting?
In a nutshell, Fairgen will reduce your niche segment's margin of error while classic reweighting techniques will not.
When applying different weights for specific segments in a quantitative surveys there is no impact on confidence intervals for that group.
