Description
Q1: Self-organizing map analysis of thermal conductivity dataset
The following code SOM-HexagonalTopology.py apply SOM unsupervised clustering to our
thermal conductivity dataset and shows the cluster map, and maps the data samples to each of
the clusters with colors showing its 1 out of 10 grades corresponding to their percentiles.
(1) Add your code from line 99, so that your SOM map can show different colors for
samples of different thermal conductivity grade (from 0 to 9, corresponding to their
percentile).
Hint: this function from scipy which can calculate the percentile of a value in a list of
numbers
from scipy import stats
print(stats.percentileofscore(target, 500))
You need to install the minisom by: pip3 install minisom
https://github.com/JustGlowing/minisom more info.
(2) If possible, try to fix the legend bar so that the color show the range of thermal
conductivity values. (optional, bonus points: 10)
Q2: Genetic programming for symbolic regression
Study this fastsr symbolic regression package
https://github.com/cfusting/fast-symbolic-regression
read the thermal_dataset.csv file, use all the numeric columns except the y-exp and y-theory
columns as the X_train, use the y-exp as the y_train
train a symbolic regression model for this dataset
print out the final regression score
print out the formula of the best individual
plot the final regression scatter plot.
