Bias in Artificial Intelligence is either embarrassing and unethical or costly or both. It can quickly occur with unbalanced samples and small samples – or Small Data.

Samples may be small because (a) nobody saw a need to collect more data in the past, (b) data collection is difficult and expensive, such as with behavioral variables in real life, or (c) the right data may be naturally sparse. For example, consider white space analysis or anomaly detection: The lack of the right data is inherent, an essential character of the phenomenon: An event is called an anomaly because its occurrence is unusual and rare (Webster 2020). With one “signal” or indicator per year, even 10 years’ worth of data represents a sample of just 10 data points or n = 10. There are just 10 signals, the rest is “noise” (Silver 2012).

Why bother? More Small Data is coming with key trends in IT, such as Edge-computing and 5G connectivity (Gartner 2019). Fortunately, there is also more research on mitigating problems with Small Data, such as:

Sikora, R., and C. Schlueter Langdon. 2019. Marketing to “Minorities”: Mitigating Class Imbalance Problems with Majority Voting Ensemble Learning. Frontiers of Marketing Data Science Journal (Fall): 27-33