Resolved: Inference x prediction in multicollinearity
Hello, regarding the video on multicollinearity, I understood that multicollinearity may not significantly impact prediction but could pose risks to inference. But what are the defitions of inference and predictions here? Is inference the prediction of unseen data?
Hi Ryo Sato,
In the context of statistics and machine learning, the terms "prediction" and "inference" have specific meanings, and they are not the same thing.
Prediction: This refers to the process of using a model to generate outputs (usually future or unknown values) based on new inputs. For example, if you have a model trained to predict house prices, using it to estimate the price of a new house based on its features (like size, location, number of bedrooms) is a prediction.
Inference: Inference, on the other hand, is about understanding the relationships between variables in your model. It's about interpreting the model to learn about the data or the phenomenon being modeled. For instance, in a model predicting house prices, you might be interested in understanding how much each feature (like location or number of bedrooms) contributes to the house price.
Hope this makes sense!