learn-co-curriculum / dsc-ridge-and-lasso-regression Goto Github PK

The following sentence appears under the heading 'Penalized Estimation':

"Now, instead of completely "deleting" certain predictors from a model (which is equal to setting coefficients equal to zero), wouldn't it be interesting // so-called penalized estimation operates in a way where parameter shrinkage effects are used to make some or all of the coefficients smaller in magnitude, closer to zero.'

I have inserted '//' to indicate where the two incomplete sentences are combined.

Link to Canvas

https://github.com/learn-co-curriculum/dsc-ridge-and-lasso-regression

Issue Subtype

• [x ] Master branch code
• Solution branch code
• Code tests
• Layout/rendering issue
• Instructions unclear
• Other (explain below)

Describe the Issue

Source

Concern

(Optional) Proposed Solution

What OS Are You Using?

• OS X
• [ x] Windows
• WSL
• Linux
• Saturn Cloud from Canvas

Any Additional Context?

the equations are broken

Link to Canvas

https://learning.flatironschool.com/courses/6223/assignments/226623?module_item_id=529510

Issue Subtype

• Master branch code
• Solution branch code
• Code tests
• Layout/rendering issue
• Instructions unclear
• Other (explain below)

Describe the Issue

Source

$$ \sum_{i=1}^n(y_i - ((m_1x_{1i})-b)^2 + \lambda m_1^2 + (m_2x_{2i})-b)^2 + \lambda m_2^2)$$

Concern

The formula is missing a parenthesis. It makes it unclear which part should be in the summation function and which part should not be summed. Also, nowhere in the lesson is it noted what p is. I'm guessing it is the number of parameters, but if so, the same variable should be used for both sums (k is used earlier). If it is different, then the text should note 'where p is ...'.

Upon another look at the formula and in an attempt to work it out myself from the base formula with k=2, I don't believe this the above correct (although there is a real possibility I'm doing my math incorrectly). See below for what I think the formula might actually be?

(Optional) Proposed Solution

$$ \sum_{i=1}^n(y_i - m_1x_{1i} - m_2x_{2i} -b)^2 + \lambda m_1^2 + \lambda m_2^2)$$

What OS Are You Using?

• OS X
• Windows
• WSL
• Linux
• Saturn Cloud from Canvas

Any Additional Context?

Let me know if I'm wrong + if so why! I'm always happy to be wrong as long as it helps me learn :)