Linear Regression with Multiple Regressors - Financial Risk Manager Video Lecture | Risk Management: An Essential Training - Management

simply learn your pace your place
financial risk manager frm part 1 the
frm exam covers the fundamental tools
and techniques used in risk management
and the theories that underlie their use
multiple regression
in the previous sessions we have learned
about regression analysis which is used
to find the linear relationship between
the variables we have assumed that the
dependent variable depended on a single
independent variable that is there was
just one regressor we obtained a linear
equation of the model in the form of y
equals B 0 plus b1x we've also seen that
this equation was not perfect in
predicting the value and that there was
an error term the effectiveness of the
model depended on how small the error
terms were we had ignored the
possibility of dependent variable
depending on other factors in this
session we will introduce the concept of
multiple regression in this model we can
have multiple independent variables that
determine the value of the dependent
variable thus we can significantly
reduce the error in predicting the value
of the dependent variable agenda we'll
begin the session by learning about the
effects of emitting the variables that
should have been included while
predicting the value of the dependent
variable we will then look at the
various assumptions in a multiple
regression most of which will be common
to our earlier assumptions made in
linear regression with one regressor we
will then analytically determine the
goodness of fit for the model through
coefficient of determination and another
measure known as adjusted r-square
finally we will look at a major source
of error in multiple regression which is
multicollinearity emitted variable bias
it is possible that a variable depends
on more than one factor for example in
an earlier session we said that petrol
prices might be a function of crude
prices and taxes as well such variables
should be included in the model for
accurate prediction if these variables
are emitted they become part of the
error term in the regression model we
also assumed in the regression model
that the error term is not car
related with the independent variable in
this case when we have a variable which
is part of the error term and also
correlated with the independent variable
this assumption will no more be valid as
now the error term will also bear some
correlation with the independent
variable as such any variable that
affects