03/08/2020 - studying

• been rather irritable and quarrelsome

• been feeling rather frustrated after socialising with friends
• it is not too good for self esteem
• not their fault, but insecurity rears its ugly head
• i don't know why i feel a need to justify myself

• not been sleeping on time
• probably Liver Fire in TCM terms

• the machine learning topics are difficult
• so it is harder to focus
• i end up experiencing regret at the end of the day
• so i stay up late trying to make up for lost time

• not been doing my youtube workouts
• so kinda frustrated also because i felt like a failure (mildly)

• i took a morning walk to the mall instead
• treated myself to mcdonald's and KOI oolong tea

1. effects of learning from videos and books 
2. after drinking from the firehose
3. my info is an unorganised mess
4. i decided to use the handwriting method
5.  scribbling handwritten notes on microsoft surface 3 using the windows ink workspace
6. it is launched from the task bar
7. and copying those to my evernote notebook
8. bit by bit i save and bank in those pieces of knowledge
9. helps clear up my thinking so i feel less frustrated and despondent

• been toggling between that and andrew ng's course

• my aim is to understand the math enough to know which methods to use for python library

machine learning - supervised and unsupervised learning

simplest algorithm - linear regression

solving systems of linear equations using matrices

• linear regression
• - python.numpy.linalg
• - solve()
• - dot()
• 

explanation of cost function and gradient descent

- to find the best fit (line)

- in calculus you can actually do differentiation or find the derivative for the minimum point

- probably gradient descent scales

linear regression cost function and optimisation with gradient descent

- MSE (check for outliers first)

- learning rate

- gradient descent (local minimum == global minimum) for lowest cost