Final Module (8, 9, 10) Flashcards
a function of random variable with a fixed distribution parameters, while Likelihood function is a function of model parameters for a fixed data set
To find the Likelihood of a data set
find the likelihood of each data point under specific parameters (pdf), and then multiply all of them (joint pdf)
- Use LogLikelihood LL=ln(L) to simplify calculations
The model with the highest LogLikelihood is
the most likely distribution for the data
Derivative
is the function that shows instantaneous rate of
change of the given function for all values of π₯
Velocity
the derivative of position with respect to time
Acceleration
is the time derivative of velocity
Steepness
derivative of height with respect to horizontal displacement
Extremum (or extreme value)
a point at which a maximum or minimum value of the function is obtained in some interval
- When π(π₯) reaches its extremum, πβ² π₯ = 0
Extrema at maximum and minimum
- At maximum: πβ² π₯ = 0, πβ(π₯) < 0
- At minimum: πβ² π₯ = 0, πβ(π₯) > 0
What if πβ(π₯) = 0?
Then itβs a point of inflection β the point where the curvature changes sign
Duration of a Bond
Cash-weighted average of time it takes to receive the bondβs cash flows, including both coupon payments and the bondβs face value
Sensitivity of a Bond
derivative of a Bond divided by y
Linear Regression
a process that determines the parameters of a linear model describing the relationship between the response variable and one or more explanatory variables
One explanatory variable
simple linear regression
Two or more explanatory variables
multiple linear regression
Normal equations
a system of equations for π½1 and π½0 after we take the derivatives of πππ with respect to π½1and π½0
πππ =
πππ = πππ + πππΈ
π ^2 equation
= πππΈ / πππ
= 1 β πππ
/ πππ
π ^2 definition
βcoefficient of determinationβ β shows what percent of variation in π¦ is explained by the linear model
- The quality of fit / coefficient of determination
Assumptions of Linear Regression
- Linear relationship between π₯1, β¦,π₯π and π¦
- The residuals β π(0, π2)
- π2 is constant across all values of Λπ¦π (homoscedasticity)
- Cov(ππ , π π )β the residuals for different data points are uncorrelated
- What if the relationship is not linear?
- Sometimes π¦ depends linearly on π(π₯1, β¦,π₯π)
Adjusted π ^2 definition
used to account for multiple independent
variables
Adjusted π ^2 equation
π πππ^2 = 1 β (1 β π ^2) * π β 1 / π β π β 1
πβ number of data points
πβ number of independent variables