John F. Elder presented the top 10 data mining mistakes at the 2005 Salford Systems Data Mining Conference. The mistakes included lacking sufficient data, focusing only on model training accuracy, relying on a single data mining technique, asking the wrong business questions, only considering the data and not domain expertise, allowing leaks from future data, discounting anomalous cases, extrapolating models too far, trying to answer every inquiry instead of acknowledging uncertainty, casual sampling methods, and believing that the single best model is correct. Elder emphasized the importance of experience, multiple techniques, asking the right questions, incorporating domain knowledge, careful sampling, and model bundling to avoid these mistakes.