Breiman and Shmueli Article Comparison

Shmueli’s “To Explain or to Predict” and Breiman’s “Statistical Modeling: The Two Cultures” articles both argue that statistics with traditional methods emphasize too much trust in familiar statistical models.

• Breiman’s focus is on the divide between data modeling and algorithmic modeling, that there’s a heavy focus on regression models and should rather focus more on predictive accuracies.

• Shmueli emphasizes that explanation and prediction aren’t the same objective, that explanatory models are utilized to test casual theories while predictive models are used to make predictions for new data. Additionally, the author states that a model can explain well but predict poorly.

In conclusion, Shmueli’s arguments helps build Breiman’s argument. Breiman proposes the challenges of the dominance of traditional statistical modeling, while Shmueli describes it. The takeaway from both articles is that the methods should be parallel to the purpose of the analysis.