Databricks Certified Professional Data Scientist Exam - Topic 6 Question 82 Discussion

Actual exam question for Databricks's Databricks Certified Professional Data Scientist exam

Question #: 82
Topic #: 6

[All Databricks Certified Professional Data Scientist Questions]

Refer to exhibit

You are asked to write a report on how specific variables impact your client's sales using a data set provided to you by the client. The data includes 15 variables that the client views as directly related to sales, and you are restricted to these variables only. After a preliminary analysis of the data, the following findings were made: 1. Multicollinearity is not an issue among the variables 2. Only three variables-A, B, and C-have significant correlation with sales You build a linear regression model on the dependent variable of sales with the independent variables of A, B, and C. The results of the regression are seen in the exhibit. You cannot request additional dat

a. what is a way that you could try to increase the R2 of the model without artificially inflating it?

ACreate clusters based on the data and use them as model inputs

BForce all 15 variables into the model as independent variables

CCreate interaction variables based only on variables A, B, and C

DBreak variables A, B, and C into their own univariate models
In statistics, linear regression is an approach for modeling the relationship between a scalar dependent variable y and one or more explanatory variables (or independent variable) denoted X. The case of one explanatory variable is called simple linear regression. For more than one explanatory variable, the process is called multiple linear regression. (This term should be distinguished from multivariate linear regression^ where multiple correlated dependent variables are predicted, rather than a single scalar variable.) In linear regression data are modeled using linear predictor functions, and unknown model parameters are estimated from the data. Such models are called linear models. Most commonly, linear regression refers to a model in which the conditional mean of y given the value of X is an affine function of X. Less commonly: linear regression could refer to a model in which the median, or some other quantile of the conditional distribution of y given X is expressed as a linear function of X. Like all forms of regression analysis, linear regression focuses on the conditional probability distribution of y given X, rather than on the joint probability distribution of y and X: which is the domain of multivariate analysis.

Show Suggested Answer

Suggested Answer: A

by Kenny at Apr 05, 2025, 12:46 PM

Limited Time Offer

25%

Off

Get Premium Databricks Certified Professional Data Scientist Questions as Interactive Web-Based Practice Test or PDF

Contribute your Thoughts:

Submit Cancel

Elvera

4 months ago

I think breaking them into univariate models could work too!

upvoted 0 times

...

Susana

4 months ago

Totally agree, forcing all 15 variables is a bad idea!

upvoted 0 times

...

Mila

4 months ago

Wait, can we really create interaction variables without more data?

upvoted 0 times

...

Herminia

4 months ago

Clustering sounds interesting, but is it really necessary?

upvoted 0 times

...

Nydia

5 months ago

Only A, B, and C matter for sales, right?

upvoted 0 times

...

Claribel

5 months ago

Breaking A, B, and C into univariate models doesn't seem like it would help with R2. I feel like we need to focus on how they interact instead.

upvoted 0 times

...

Carolynn

5 months ago

I think creating interaction variables based on A, B, and C could be a solid option. We practiced a similar question where interaction terms improved the model fit.

upvoted 0 times

...

Ty

5 months ago

I'm a bit unsure about the best approach here. Forcing all 15 variables into the model seems risky, but I think creating clusters could also complicate things.

upvoted 0 times

...

Silvana

5 months ago

I remember we discussed how interaction variables can sometimes help capture relationships that aren't obvious with just the main effects. That might be a good way to increase R2 without adding noise.

upvoted 0 times

...

Elbert

6 months ago

This seems pretty straightforward to me. I'd just go with option C and create the interaction variables. That's a solid way to try to improve the model without violating any of the restrictions. As long as the interactions are meaningful, that should give us a nice boost in R-squared.

upvoted 0 times

...

Gerald

6 months ago

Okay, I think I've got a strategy here. Since we can't add more variables, I would focus on creating interaction terms between A, B, and C. That could help capture some of the more complex relationships that might be driving sales, without artificially inflating the R-squared.

upvoted 0 times

...

Alaine

6 months ago

I'm a bit confused by the restriction to only using the 15 variables provided. Wouldn't it be better to try adding some additional variables that could be relevant, even if they aren't directly correlated with sales? That might help increase the model's explanatory power.

upvoted 0 times

...

Socorro

6 months ago

Hmm, this is an interesting question. I think I would start by looking at the variables A, B, and C in more detail to see if there are any interactions or nonlinear relationships that could be captured to improve the model's R-squared.

upvoted 0 times

...

Vivan

10 months ago

Why don't we just throw in the kitchen sink and see what happens? I bet that would really boost the R2. What could go wrong?

upvoted 0 times

...

Dorethea

10 months ago

I'm with Hildegarde on the interaction variables. That's the way to go if we want to get the most out of the data without cheating.

upvoted 0 times

...

Leonora

10 months ago

Forcing all 15 variables into the model is a bad idea. That's just going to lead to overfitting and won't give us any meaningful insights.

upvoted 0 times

...

Eden

11 months ago

Breaking the variables into their own univariate models doesn't seem like it would improve the overall R2. We need to look at the combined impact of the key variables.

upvoted 0 times

Billy

10 months ago

Breaking the variables into their own univariate models doesn't seem like it would improve the overall R2. We need to look at the combined impact of the key variables.

upvoted 0 times

...

Ramonita

10 months ago

C) Create interaction variables based only on variables A, B, and C

upvoted 0 times

...

Tamie

10 months ago

A) Create clusters based on the data and use them as model inputs

upvoted 0 times

...

Destiny

12 months ago

That's an interesting point, but I still think breaking variables A, B, and C into their own univariate models could also be a valid strategy.

upvoted 0 times

...

Hildegarde

12 months ago

I think creating interaction variables based on A, B, and C could be a good way to increase the R2 without artificially inflating it. That would allow us to capture any non-linear relationships between the variables.

upvoted 0 times