Classical regression model
Linear regression model
Linear regression model is identifying the regressions between this part, and how they affect a back-end actiity
Model with no interactions VS Model with interactions
no interactions : parallel
interactions: don’t need to be parallel, it allows to interact
How does deep learning capture interactions
可视化，来自真实世界的data:
Input layer: this represents our predictive features like age or income.
Output layer: the prediction of the model
无法从真实世界直接观察到的data:
Hidden layer: all layers that are not input layer or output layer
- each node of a hidden layer is a aggregation of information from our input data and enable the model to capture interactions.
- The more nodes we have, the more interactions we can capture.