Normalization / Standardization

mathewmanoj.wordpress.com/mul...

[@rushdierasheed6662]

Dear Mr Mathew, I am an MSc student and have been struggling for over 2 months with my research methodology and research design. I should say that after reading a number of books and research papers I was lost more and more, then I accidentally came across your video. Your explanation is crystal clear and to the point. What took me 2 months to understand only took me 30 min from your video. I strongly believe that teaching is a unique talalent, which very few people like you are gifted. Thank you for this series of videos and god bless you for your efforts...!!!

[@amankumar-kw3xs]

The best reviewer comment ❤️

[@ankitasingh2254]

Sir as there are three methods of MCDM
1. Boolean (OR , AND )
2. Weighted linear Combination
3. Ordered Weighted Average
I request you sir to make a video on these concepts too!
It will be a great help to all the students

[@jintailiu7846]

Thank you for the excellent explanations! They are so vivid with your tables!

[@yulinliu850]

Thank you so much for this excellent video! I feel, linear normalization (sum) could be quite useful for evaluating alternatives against quantitative criteria in an AHP model.

[@pablocesarmanyomavelasquez510]

Very clear, thank you Manoj

[@Leonardo-jv1ls]

Thank you a lot. Amazing video.

[@duetplay4551]

thx a lot, sir. Your teaching is easy but has depth.

[@mattglinsmann2233]

This was great! possibly using different normalization’s for different columns might be useful in some situations

[@nidhalbaccouri1799]

Excellent Explanation. Thanks, I have a question maybe someone here can answer me. If I'm using neural network and need to normalize data, which way would be the best for using Linear Regression and which would be better for Classification.

[@ravindrasinghsaluja2905]

In given normalization techniques separate equation for cost and benefit criteria means after normalization all criteria value changes to higher the better like benefit criteria.

[@simardeep9286]

if we perform feature scaling and PCA also, then which one should be applied first? if we firstly normalize and then perform PCA that reduces dimensions then why to normaliize that dimesions that would be reduced

[@kasungimhana2841]

Hi all, can somebody explain me which normalization method will suit me, to select the best intervention for municipal assets considering the cost and time as criteria? Thanks in advance.

[@SSGSolarSystemGaming]

Most of the time I saw, linear normalisation (sum) is used in AHP, vector normalisation is used in TOPSIS and linear normalisation (ii) is used in SAW method.....and in your video I saw in case of CRITIC weighting method you are using linear normalisation (max-min).......what is the reason behind to this of using different normalisation method in different MCDM methods??.....

[@RAM-xr5ve]

very nice sir

[@sophia2928]

thank you :)

[@guraysonugur]

Thank you very much

[@akibjavedr]

Thanks for the helpful video. Could you add MAD normalization here?

[@devakrishnan9439]

Which type of normalization method using are which type of MADM problems.

[@ansarshaik8908]

sir plz.....
explain the CRITIC for weight determination of criteria

[@vinr1187]

Hi Manoj = How do we normalize qualitative data ?

[@jamescreativity]

For the non-beneficial min-max method, would (Xij - MAX) / (MIN - MAX) not work better?

[@bojanbatinic4591]

Hi, thank you very much for this video - it was very useful! I wonder did you maybe make mistake in Logarithmic Normalization, since you did logaritham of each value and than their Product, while in formula is first to make Product of all values and than logaritham of that Product? Thanks!

[@salmanahmad433]

I agree with Bojan. You are getting 8.53 while according to formula value should be 22.1 for cost criterion.

[@suranjanachaudhuri340]

same observation. I back-calculated the formula used. Now the question is, which one is correct?

[@therezarakelcunha6777]

Is there any difference in results choosing different methods of normalization?

[@manojmathew5287]

Yes the final results (rank) may vary with different methods

[@shaiksubhani3247]

the normalization values obtained here are different in various normalization methods used, is it correct
whatever normalization method employed the values have to be same , but in your considered examples ,the normalization values are very different

[@manojmathew5287]

The methods are correct, the meaning of normalisation is converting all values on same scale. i.e. values between 0 to 1. Its upto the decision maker, the type of normalisation he want to use.

[@nagandranathvemishetti9247]

can u provide the code for implementing in python

[@zainabkhan2475]

1+ subscriber

[@MTauhaA]

The normalization methods that result in 0 aren't suitable for entropy method. Also, the logarithmic normalization formula for non-beneficial criteria is not correct. the big term on RHS is not subtracted by 1.

[@kashifqureshi5581]

Standrization??

[@JyotiSharma-yx5ux]

Very well explained Sir. I am comparing performance ratios of 5 banks for 10 years. I have normalized each ratio for each bank using linear normalisation(max-min) method and I get values between 0 and 1. Now I want to compare these normalised ratios of 5 banks using MCDM. However, now I am finding it difficult to normalise these already normalised ratios as their values are between 0 and 1. Can you please advise how it should be done? Thanks.

[@manojmathew5287]

Normalisation is done to convert values of criteria in the range of zero to one. So if you already have the values in the mentioned range, then no need of performing normalisation again.

[@JyotiSharma-yx5ux]

@@manojmathew5287 Thank you Sir for the clarification.

[@JyotiSharma-yx5ux]

@@manojmathew5287 Thank you Sir. Really appreciate.

[@kunalkiran3318]

Sir you performed log on individual values ; then took their product; then again performed log on the product. This gave you the denominator needed.
So basically, log of product of log values of individual x(ij).
While in formula it says, log of product of x(ij) and not log of product of LOG of x(ij).
Sir, please let us know which one is the correct one.

[@manojmathew5287]

Sorry for the mistake in the calculation. The formula shown in the video is correct.

[@ebnerashid3702]

What to do if single data of any alternative is not available?? Say, Storage of mobile 3 is not available but all others are available.
Your reply will help me a lot. Thank you

[@manojmathew5287]

Remove that criteria or remove that alternative. That's called data filtration.

[@ebnerashid3702]

@@manojmathew5287 thank you very much sir

[@mazenfarid3231]

can i change between beneficial and non beneficial formula in normalization?

[@mazenfarid3231]

i mean in linear normalization.

[@suranjanachaudhuri340]

@@mazenfarid3231 in this case, it will be against human reasoning. Otherwise, no problem

[@gulsenkeskin1303]

TODIM desktop application installation file link:github.com/gulsenkeskin/TODIM-Masaustu-Uygulama
I would be very pleased if you share your opinions. Thanks