Chapter 12 Project: Regression Of Boston House Prices

Question 1

How can we create a kerasregressor model with a standard scaler? P 87

Answer 1

import numpy
import pandas
from keras.models import Sequential
from keras.layers import Dense
from keras.utils import np_utils
from sklearn.model_selection import cross_val_score
from sklearn.model_selection import KFold
from sklearn.preprocessing import LabelEncoder
from sklearn.pipeline import Pipeline
from sklearn.datasets import load_breast_cancer, load_iris,fetch_california_housing, fetch_openml
from sklearn.preprocessing import StandardScaler
dataset =  fetch_california_housing(as_frame=True)
X = dataset.data
Y = dataset.target

def baseline_model():
  model=Sequential()
  model.add(Dense(8,input_dim=8,activation='relu'))
  model.add(Dense(1))
  return model
estimator=Pipeline(steps=[('scaler',StandardScaler()),('model',KerasRegressor(model=baseline_model, epochs=10, batch_size=100, verbose=0,random_state=0,loss= "mean_squared_error",optimizer= "adam"))])
kfold = KFold(n_splits=3)
results = cross_val_score(estimator, X, Y, cv=kfold,scoring='neg_mean_squared_error')
print("Baseline:  %.2f (%.2f) MSE" % (results.mean(), results.std())) #MSE is negative when returned by cross_val_score

Question 2

One way to improve the performance of a neural network is to add more layers. This might allow the model to extract and recombine higher order features embedded in the data. True/False? P 92

Answer 2

True

Question 3

Another approach to increasing the representational capacity of the model is to create a wider network. True/False? P 93

Answer 3

True. For example: Keeping a shallow network architecture and nearly doubling the number of neurons in the one hidden layer can help

Question 4

It would be hard to guess that a wider network would outperform a deeper network on a problem. This demonstrates the importance of empirical testing when it comes to developing neural network models. True/False P 94

Answer 4

True