machine learning

Question 1

model

Answer 1

takes observations, makes prediction

use ML to generate model

Question 2

Supervised Regression Learning

Answer 2

S= Tell system correct output for given input
R= Numerical approximation
L= Train with Data

Question 3

Classification Learning

Answer 3

Classify

Question 4

Linear Reggression

Answer 4

Find parameters for model (parametric learning)

Question 5

K Nearest Neighbor (KNN)

Answer 5

Make prediction based on previous x/y data sets

Question 6

Decision Tree

Answer 6

Each tree node is a decision, leads to leaf that is the model

Question 7

Decision Forests

Answer 7

Combined Decision Trees

Question 8

Supervised Regression LearningTypes

Answer 8

Linear Regression
K Nearest Neighbor
Decision Tree
Decision Forest

Question 9

Basic ML Steps

Answer 9

Select Predictive Factors
Select Output (generally future price)
Select Time Period
Select Stocks
Train Model

Question 10

Backtesting

Answer 10

Use set of training data to generate model

See how well model compares to reality

Question 11

Regression Learning Problems for Trading

Answer 11

Noisy and uncertain
Challenging to estimate confidence
How long to hold position
How much to allocate to position

Question 12

How to address issues

Answer 12

Use Reinforcement Learning to teach policies

Question 13

Parametric Model

Answer 13

Apply regression line to model
linear : y=mx+b
polynomial: y= mx^2+mx+b

Question 14

KNN Application

Answer 14

Take average of nearest K samples to determine output

Question 15

Kernel Regression

Answer 15

Weight contributions of each K data point based on distance

Question 16

Out of Sample Testing

Answer 16

Seperating Training Data from Testing Data