Answer-first summary for fast verification
Answer: Creating a feature cross of latitude and longitude and bucketizing it at the minute level while using L1 regularization during optimization
L1 regularization is preferred when you need to emphasize more influential features by shrinking less important features to 0. This approach is particularly useful when dealing with location data, as it allows the model to focus on the most relevant geographic features for predicting housing prices. L2 regularization, on the other hand, is more suitable when all input features have a similar influence on the output. Reference: [Google Machine Learning Crash Course](https://developers.google.com/machine-learning/crash-course/regularization-for-sparsity/l1-regularization)
Author: LeetQuiz Editorial Team
Ultimate access to all questions.
When training a neural network to predict housing prices using a dataset that includes latitude and longitude as features, what is the most effective method to account for the physical dependency of location on price?
A
Creating a numeric column from a feature cross of latitude and longitude
B
Providing latitude and longitude as input vectors to the neural net
C
Creating a feature cross of latitude and longitude and bucketizing it at the minute level while using L1 regularization during optimization
D
Creating a feature cross of latitude and longitude and bucketizing it at the minute level while using L2 regularization during optimization
No comments yet.