Competition of Week 1 Statistical Modelling 05/02/2026 and HomeWork file for week 1

[MarcoRianiUNIPR]

The homework file can be downloaded here

wafer.pdf

Solution will be discussed during next class.

The first who obtains the plot below will get +1

[BoschiGiacomo]

This is my solution

clear; close all;

% load the data and add the rownames as required

% XrowNam= "w" + (1:10);
% X= readtable("wafer.txt", "ReadVariableNames", true);
% X.Properties.RowNames=XrowNam;

rnames= "w" + (1:10)';
X= readtable("wafer.txt", "ReadVariableNames", true);
Xt= addvars(X, rnames, 'Before',1);
%% Parallel plot
parallelplot(Xt);

To obtain the plot you created, i had to add the variable names as a variable, but i also left my previous solution to the exercise.

This is the result of the code:

[MarcoRianiUNIPR]

Absolutely correct! Well done and congratulations @BoschiGiacomo

Small remarks to your code

"ReadVariableNames", true

is unnecessary because it is the default

The second remark is that if your input txt file changes and additional rows are added this will not work anymore because you used

rnames= "w" + (1:10)';

[BoschiGiacomo]

Thanks for the suggestions! I'm used to always leaving it in for the sake of clarity, but i agree that it's unnecessary.
In hindsight, i should have used the height of the table as an upper limit to make the code reusable, i overlooked that.

[missblissgraphics]

T=readtable("wafer.txt") ;
T.Properties.RowNames="w"+(1:height(T)); %adds Row Names

WaferType=T.Properties.RowNames;

T = addvars(T, WaferType, 'Before', 1); %adds Row names as first column
parallelplot(T); %2d plot for all variables in table T

[MarcoRianiUNIPR]

@missblissgraphics it is not clear what your solution adds to the solution of @BoschiGiacomo 2 days ago.
Please do NOT post code duplications!

[missblissgraphics]

@MarcoRianiUNIPR Understood Sir. I had a question though, even though the range differs for all 3 variables, would using zscores be a bad idea here?

[MarcoRianiUNIPR]

Of course zscores can always be used.
However, given that function parallelplot automatically adds the correponding axis of the variable, it is unnecessary.
For example the code

load citiesItaly.mat
parallelplot(citiesItaly)

produces

For example the fact that unemploy is "strongly" negatively correlated with pensions and export can immediately be seen, even if the the variables are not standardized.
Similarly, for example, the positive correlation between addedval and deposit can immediately be seen.

In summary, if I have to use boxplots to compare the distributions of the variables it is necessary to preliminarly standardize the data. With parallel coordinates this is unnecessary.

However, using zscores is always fine.
Option DataNomalization of parallelplot enables us to specify how the data must be standardized

https://www.mathworks.com/help/releases/R2025b/stats/parallelcoords.html

Please let me if I have answered your question.

[missblissgraphics]

Yes I understood. Thankyou.

[MarcoRianiUNIPR]

Here is the solution, feel free to ask everything which is unclear!

waferOUT.pdf