What is the best Data Mining algorithm for predicting based on a single variable?

I have one variable whose value I would like to predict, and I would like to use only one variable as predictor (for instance: predict traffic density based on weather). Initially I thought about using Self-Organizing Maps, which performs unsupervised clustering + regression, but since it has an important component of dimensionality reduction I see it as more appropriated for a large number of variables. Does it make sense to use it for a single variable as predictor? Maybe there are more adequate techniques for this simple case: I used "Data Mining" instead of "machine learning" in the title of my question, because I think maybe a linear regression could do the job...