Bias and variance are two major issues we try to manage in machine learning. We are always looking for a trade-off between the two. Bias has to do with the training data while variance has to do with the model output. Bias occurs when your training data does not adequately honour the underlying assumption upon which the problem to be solved is based. Variance occurs when your model predicts away from the ground truth (expectation).
How do you handle bias in your model design? How do you compose your training data? Do you combine measurements from different vendors or service companies to build training database? I strongly recommend this article for machine learning enthusiasts and practitioners. Let me know what you think about it.