# How ML models can be optimized for more dimensions?

If you do this so that you remember these concepts RNC will never remember these concepts Familiar It is most important So I kindly request you least open Except those off You have not opened Accelerated low Only when you’re putting Only when we put in some effort we will try to remember If you don’t put any effort will never want to remember those things 3.611 This value is nothing by my Eigenvalue for the faster dimension What is the I can value there for the first dimension 3.6 Now Similarly if you calculate for the second column it will be equal to 1.8 Will it not be equal to one Find eight.

Do it and stick to it and see plastic If it’s working out correctly Squire distance Add them Tell me what is the total value 1.85 So that is what is plotted here Character And so 3rd 1 when you calculate it will be close to 1.191 points two right This one is very close to 0.2 or zero points three or whatever Be right So I again value a calculator for all the seven dimensions So if you look at the whole table the other four dimensions as well you calculate Squire them calculates choir them calculator You will have the Eigenvalues for them so automatically what is happening here My faster dimension is that dimension which has the I highest I e in value All right So my dimensional.

The election is happening In what manner The first dimension is nothing but that dimension which has the highest Eigenvalue meaning that diamond chain which can explain maximum variance off all the very was put together clear days That dimension which can explain the maximum variance of all the very was put together that this considered as of May first principle component right or not again value We’ll lose the magnitude of the ah particular competent principal company Is it clear So nothing bad for the 1st 2 principal conferences Quieting these for option values is what we’re doing We are squiring this question values And once we do that we get the magnitude of this particularly in every question That is what principal component is salable PCs I see both factor analysis and principal competent answers give you the same old put Are you getting it.

They bought to give you the same output The only difference between offend PCs affair takes into consideration What is the community effect for every variable BC a blindly does first competent Okay is the highest I can really What is there But it was the first company blindly take what is the next principal competent which has the next highest Eigenvalue on that list The same When you look at the output of board the first confident that is derailed Far factory Lance’s RPC will be the same It will be the same but in the output, you will have the uniqueness listed for you for factor analysis meaning you will be able to understand how much of every variable is explained within this tree in the principal component analysis I will not look into that are getting.

If you want to understand when you’re reducing 10 variables into three variables are three dimensions are four dimensions If you want to understand how much of each variable is being explained within these three or four dimensions you go for factor analysis Yeah you feel if you do not get about it go hard Principal confident finances the same example that I gave you If you understand your data set to go for factor analysis If you don’t understand their data set to go for PC pc a Does not give you the insight Affair gives you the insight as to each other It will hike height I’m able to see that 96% off height variables are actually included within the 1st 3 components right So factory answers will help me show these values Understand these values principal competent answers will not help you in this It will only say this money very ones that you gave me These are all the principal components No Choose whichever conference that you want to choose.

I’m go ahead with you If we start doing that I think the next one never will go in calculations Only way we do it in our directly Yeah I can give you an excel where in each and every values calculated I can give you the farm last after 10 except except but in class 31 way to do the farm last itself Right What I’m explaining to you please note it down in the notebook I’m giving you all the calculations right Squire Each and every values these you have right What is it that you want That has a very complex structurally question and bottom line which is which is included in what we actually do is we said these are the variables that I have run principal component analysis are we’ll check the output for me automatically saying these are your noting values as simplistic I will give you the open sea This is nothing but this is our output I’m showing you the output in our the clear guys We are not going to calculate this matter Maggie We are not going to calculate commonalities mathematically getting it done We are not going to add subtract multiply do anything.

We’re only going to say a principal company I will call the function of principle compromises our factory ounces give the input variables and everything else is done for us In some breakage, there is a small weightage which is either to Ali variables Yes but that but the wait are correlation between every variable and the competent itself The first are you getting it So that is what I’m saying when you Squire and add them across the square and add them across all the seven will give you a dork Love one commonality if one meaning the entire variable is explained by all the seven dimensions put together are getting it But the fastest three can explain 96% off it The other four is explaining only 4% off it 40.4 off it and We see someone is the worst.

A competent will not take the competent at all How do you get in here BC seven will actually be explaining very close toe around the half a percent of your variables each Are you getting it because it is always in ascending order The best to complement this year faster competent because it has your highest Eigenvalue the Second complaint would be that next to piss company out of your seven components so it’ll always be shown in descending order only So if there is weak correlation if there is we correlation between all your variables all the seven components we need to be taken into account which makes no sense in dimensionality reduction correct or not Instead of using seven dimensions in place of seven variables.

I will use a seven very was itself It is not the best feeling It is the It is not a line It is not the best fit line or anything It is that best dimension It is a diamond chain which explains all these variables in this particular correlation are getting it now We’ll quickly really quickly Some parts from the gun here guests We’ll proceed Yes imagine doing the same for every dimension This is also principle Competent else’s CNPC A Days Inn PCF everything you’ll get factor loadings as the output Are you getting it The approaches only different in what senses factor analysis with a commonality and show you What is the commonality Principal comment will not tell you what is a community.