A violation in GM assumptions can have business consequences because it will result in algorithmic bias. Imagine that in a hedge fund, as a data scientist, you are tasked with determining what dictates Airbnb overnight rental prices: Property type? The number of people it can accommodate? Distance of the property to city center? Cancellation policy? Having a wrong answer to that question may result in the fund purchasing properties that will not return maximum profit. To answer the question you can fit a Linear Regression model that predicts the prices and then perform a feature importance analysis. The result of the analysis may tell you that the most important features are, for example, the number of bathrooms and distance. The results of such analysis should be trusted only if the underlying assumptions are satisfied. For a more detailed discussion on this, refer to https://github.com/mtorabirad/PricePrediction/blob/master/project_predicting_AirBnB_prices.md

For more information on how these assumptions can be tested in Python, you can refer to. https://www.statsmodels.org/dev/examples/notebooks/generated/regression_diagnostics.html https://jeffmacaluso.github.io/post/LinearRegressionAssumptions/