MENU
A machine learning engineer has developed a model and registered it using the FeatureStoreClient fs. The model has model URI model_uri. The engineer now needs to perform batch inference on customer-level Spark DataFrame spark_df, but it is missing a few of the static features that were used when training the model. The customer_id column is the primary key of spark_df and the training set used when training and logging the model.
Which of the following code blocks can be used to compute predictions for spark_df when the missing feature values can be found in the Feature Store by searching for features by customer_id?
Suggested Answer: E
Get Premium Databricks Machine Learning Professional Questions
as Interactive Web-Based Practice Test or PDF
Real Exam Format
Detailed Explanations
Instant Access
Get Your Practice Test
17 comments · newest first
SK
Skye
10 months ago
E looks good, but I'm not sure if it handles the missing data properly.
VA
Val
11 months ago
D is definitely wrong, it doesn't address the missing features.
AL
Alida
11 months ago
Wait, can we really use C? It looks like it calls get_missing_features twice.
FI
Filiberto
11 months ago
I think B is too simple, it won't work without the missing features.
EL
Elouise
11 months ago
Option A seems right, it fetches missing features first.
JE
Jesus
12 months ago
I lean towards option E because it directly scores the DataFrame, but I'm not completely confident about whether it handles missing features correctly.
KR
Kristofer
12 months ago
I feel like we practiced something similar, but I can't recall if `fs.score_model` or `fs.score_batch` is the right function to use here.
ZE
Zena
12 months ago
I think option A sounds familiar because it mentions getting missing features first, which seems like a logical step.
PH
Phillip
1 year ago
I remember we discussed how to handle missing features in our training sessions, but I'm not sure if we should use `get_missing_features` before scoring.
OM
Omega
1 year ago
This is a tricky one. I'm not sure if I should be using `spark_df` or the DataFrame returned from `fs.get_missing_features()`. I'll need to think this through carefully before selecting an answer.
LE
Lettie
1 year ago
Okay, I've got a plan. I'll use `fs.get_missing_features()` to fetch the missing features from the Feature Store, then pass that DataFrame to `fs.score_batch()` to get the predictions. Seems like the most robust way to handle this.
AL
Alaine
1 year ago
Hmm, I'm a bit confused about the difference between `fs.score_model()` and `fs.score_batch()`. I'll need to double-check the documentation to make sure I understand which one is appropriate for this scenario.
FR
Frank
1 year ago
This question seems straightforward. I think I'll go with option C - it looks like the most comprehensive approach to handling the missing features.
GA
Gabriele
1 year ago
I think option C is the way to go. Fetching the missing features and then scoring the batch seems like the most complete solution to the problem. I'm feeling confident about this one.
GE
Georgeanna
1 year ago
Option C sounds promising - attaching QoS settings directly to device profiles could be an efficient way to prioritize voice traffic.
MO
Mozell
1 year ago
I'm a bit stuck on this one. The question is asking about the relationship between vendor-specific security and service autonomy, but I'm not totally clear on the implications. I'll have to review my notes.
EL
Elouise
1 year ago
I think it makes sense to add it in the Primary zone since it's where the registration failures happened. That's what we practiced in our last session.