The article presents a method to detect outliers in sparse geotechnical data by using Bayesian learning. Outliers are data points that don't fit the expected pattern. The method calculates the probability of each data point being an outlier and identifies outliers with high probabilities. By using a re-sampling technique and Bayesian learning, the method can handle statistical uncertainties caused by outliers. The study shows that this approach effectively identifies outliers in geotechnical data, helping to improve the accuracy of statistical analysis and uncertainty quantification in geoscience.