Regarding the terms of fitrsvm Algorithm

Dear All,

I am currently looking at fitrsvm algorithm. I could not make sure about some terms written on the formulas of this algorithm . Although I have glanced to some articles to learn terms on formulas, I could not obtain reasonable answer for them.

If possible, could you please support me about this situation?

Thanks in advance.

Best regards,

Answer 1

[Vatsal]

Hi,

I understand that you want to understand the terms written on the formulas of ‘fitrsvm’ algorithm.

The equation ""​​ is the formula for standardizing a predictor.

• is the standardized value of predictor j.
• is the original value of predictor j.
• ​ is the weighted mean of predictor j. It is calculated as the sum of the products of the weights and the corresponding predictor values for all observations, divided by the sum of all weights.
• is the weighted standard deviation of predictor j. It is calculated as the square root of the weighted mean of the squared deviations of predictor j’s values from their mean value.

The equation "" is the formula for calculating the weighted mean of a predictor.

• ​ is the weighted mean of predictor j.
• ​ is the weight of observation k.
• ​ is the value of predictor j for observation k.
• The sum ​ is the sum of the products of the weights and the corresponding predictor values for all observations.
• The sumis the sum of all weights.

I hope this helps!