Assignment 6#
In this week’s lab we will develop a neural network to predict house prices. The data can be downloaded from here and here.
There are two csv files. The first, house-features.csv, contains eight attributes for 20640 houses. Note that each row represents some geographic unit (e.g. Census tract) hence columns such as AveOccup are decimal numbers. The other file, house-prices.csv contains the house prices.
Task 1 (10 points):#
Develop a multilayer perceptron that predicts house prices using the features available.
To earn full marks on this assignment, we would like to see some evidence of a systematic search for model performance.
Unlike the previous assignment, this search should focus on the strategies for optimizing the model itself. Things that could be adjusted include:
Number of hidden layers
Number of nodes in the hidden layers
Learning rate
Regularization (e.g. dropout, L2 penalties)
Epochs
Batch size
Activation functions
Don’t attempt to optimize every one of these but please try at least one. You can use learning curves as evidence for this optimization search.
Note
The best performing model will be the one that can generalize to new data, so we will be checking that the dataset was split into training (80%) and testing data (20%) and that the testing error was evaluated.
Important
Please submit your notebook in .pdf format to Canvas by the deadline as evidence of your work.