includes interpretation of singularity and avoiding dummy variable trap in regression analysis. Let me know if anything is required. Available at google #bobrupakroy

1. Machine
Learning - I
Regression Analysis –
Handling Singularity Issue

2. Interpreting the Singularity Issue
In the output of regression the model infers
 If the Gender variable goes up from F to M then the purchase will go
up with an average of 496 units else if the unit goes up from M to F
then the purchase will go down with average of -496 units i.e. they
have inverse relationship.
 In the 2nd output from the derived variable GenderM as 1and
GenderF as 1, we can observe if we take only GenderF then it gives
 a negative impact in the
 purchase. Hence inverse
 inverse relationship with
 Gender Male as described
 in the above point.


3. Singularity
 Now the next variable having Age1Young, Age1Midage and
Age1Old the model says with respect to Age1Midage the purchase
will go up by average 180 units if any Age goes up from
Age1Midage to Age1old and vice versa for Age1young by -273
units.
Again using derived variable
Age1_Midage1 we can clearly
see there is a inverse relationship
with Age1Young values.
So the above statement is True if
the age goes up from Age1Midage
to Age1Young the purchase will
go down by273 units.

4. Interpreting Singularity issue
 Sometimes we might encounter estimates as ‘NA’.
For example consider the derive variable GenderM & GenderF, when
we use both either one of them will show as NA. This point of estimate is
known as singularity that means the variables are not linearly
independent in other words it is highly correlated. Because the
information given by the
variable GenderF already
contained in the
GenderM. Thus redundant.
 Hence we have seen in
our previous slide it has an
inverse relationship.

5. Next
We will learn an another type of regression known as LOGISTIC
REGRESSION if we have our target variable is BINARY.