Summary Results 141a

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 1

One of the statistical questions of interest is to review whether there are any relationships

between variables of the client’s basic information and the term deposit?

Statistic analysis
For this question, we will use the lasso and ridge models to determine whether there are any
relationships between the basic information and term deposit.
Through the lasso and ridge model, we will be able to determine any relationships between the
variables and graph out the L1 norm such that we will be able to visualise the coefficients
through the vector of coefficients. We will also graph out the CV ridge and lasso to be able to
compare and visualise the mean square error with log.
In the beginning of the investigation, we will create two versions of the data set, with 10 cross
validations. The 10 cross validations are done through each variables and their subsets. We will
then conduct the train test split, creating the training data and the testing data.
To prevent skewing of the data, we will rescale the data so the distance measures are not
skewed through scaling differences between variables.
Finally, we will run the ridge regression, with alpha =0 and noting that the family for running the
model is binomial.
Then, we will tune the model/s regularisation parameter, as well as show the model
performance metrics. This will show us the results of the model and the success rates of the
model.
Finally, we will plot the cv ridge as well as the l1 norm, this will present the mean square error
and vector of coefficients.

Results:
Through the lasso and ridge model- they both present better accuracy than the probability of
finding correct yes with no model. There is a higher correct rate with model than without.
As shown through the model predictions and coefficients run.

You might also like