Machine Learning: 2 Books in 1: Machine Learning for Beginners, Machine Learning Mathematics. An Introduction Guide to Understand Data Science Through the Business Application

"ElasticNet Regression". Its primary objective is to further enhance the
accuracy of the predictions generated by the "LASSO regression"
technique. “ElasticNet Regression" is a confluence of both "LASSO" and
"ridge regression" techniques of rewarding smaller coefficient values. All
three of these designs are available in the R and Python “Glmnet suite”.
"Bayesian regression" models are useful when there is a lack of sufficient
data or available data has poor distribution. These regression models are
developed based on probability distributions rather than data points,
meaning the resulting chart will appear as a bell curve depicting the
variance with the most frequently occurring values in the center of the
curve. The dependent variable "Y" in "Bayesian regression" is not valuation
but a probability. Instead of predicting a value, we try to estimate the
probability of an occurrence. This is regarded as "frequentist statistics", and
this sort of statistics is built on the "Bayes theorem". “Frequentist
statistics" hypothesize if an event is going to occur and the probability of it
occurring again in the future.
"Conditional probability" is integral to the concept of “frequentist
statistics”. Conditional probability pertains to the events whose results are
dependent on one another. Events can also be conditional, which means the
preceding event can potentially alter the probability of the next event.
Assume you have a box of M&Ms and you want to understand the
probability of withdrawing distinct colors of the M&Ms from the bag. If
you have a set of 3 yellow M&Ms and 3 blue M&Ms, and on your first
draw, you get a blue M&M, then with your next draw from the box, the
probability of taking out a blue M&M will be lower than the first draw. This
is a classic example of "conditional probability". On the other hand, an

---

independent event is flipping of a coin, meaning the preceding coin
flip doesn't alter the probability of the next flip of the coin.
Therefore, a coin flip is not an example of "conditional probability".