辅导IE 7275、讲解Data Mining、辅导R设计、讲解R
解析Haskell程序|调试Matlab程序
Homework 6
IE 7275 Data Mining in Engineering
Note: Read the following literature before you attempt to solve the problems in this
homework.
Neural network related documents
Documentation on neuralnet.pdf
Examples on neuralnet.pdf
Reference manual on neuralnet.pdf
Documentation on rsnns.pdf (focus on mlp)
Reference manual on rsnns.pdf (focus on mlp)
Reference manual on nnet.pdf
Visualizing neural networks in R.pdf
Using neural networks for credit scoring.pdf
Lift charts related documents
Tutorial on lift charts with R.pdf
Documentation on lift.pdf
Lift charts to compare binary predictive models.pdf
Documentation on gain.pdf
Problem 1 (Car Sales, Neural Networks)
Consider the data on used cars given in ToyotaCorolla.xlsx. The data has 1436
records and details on 38 attributes, including Price, Age, KM, HP, and other
specifications. The goal is to predict the price of a used Toyota Corolla based on its
specifications using a multilayer neural network. Select appropriate predictor variables.
Use 75% of the data for training a multilayer neural network and 25% to validate the
network performance. Use the default algorithm (“rprop+” in the neuralnet package).
Record the RMS error for the training data and the validation data. Repeat the process,
changing the threshold values, 1, 0.1, 0.05, 0.01, 0.005, 0.001, and 0.0001. Set threshold
to these values.
(a) What happens to the RMS error (or Sum of Squares Error) for the training data as
the value of threshold decreases?(b) What happens to the RMS error Sum of Squares Error for the validation data?
(c) Conduct an experiment to assess the effect of changing the number of hidden
layer nodes (default 1), e.g., 1,2,4,8.
(d) Conduct a similar experiment to assess the effect of changing the number of
layers from 1 to 2 in the network.
(e) Study the effect of gradient descent step size (learningrate) on the training
process and the network performance.
Files Included in the Folder:
1. Homework.pdf
2. Documentation on neuralnet.pdf
3. Examples on neuralnet.pdf
4. Reference manual on neuralnet.pdf
5. Documentation on rsnns.pdf (focus on mlp)
6. Reference manual on rsnns.pdf (focus on mlp)
7. Reference manual on nnet.pdf
8. Visualizing neural networks in R.pdf
9. Using neural networks for credit scoring.pdf
10. Tutorial on lift charts with R.pdf
11. Documentation on lift.pdf
12. Lift charts to compare binary predictive models.pdf
13. Documentation on gain.pdf
14. Toyta Corolla.xlsx