CertLumen
Databricks · Associate

Databricks Machine Learning Associate

Explore data, engineer features, train, tune, track and deploy models on Databricks. Lessons, exam-style practice, timed mocks and an AI tutor, mapped to every objective of the official exam guide.

Start your free trialTake the free 10-question check

₹399 12-month access · 7-day free trial

17lessons
270exam-style questions
4practice exams
6hands-on labs

The exam

Questions
48
Time
90 minutes
Passing score
Not published by Databricks; this course targets ≥ 80% on unseen mocks
Exam fee
USD 200 (plus applicable taxes)
Valid for
2 years (recertify by taking the current live exam)
Delivery
Online proctored or test center

Official exam page · booked with Databricks

Syllabus

Every section of the exam guide

4 sections and 48 objectives, in the official weights. Each lesson is mapped to the objectives it covers.

Section 1 · Databricks Machine Learning38%

What the exam asks

  • Identify the best practices of an MLOps strategy
  • Identify the advantages of using ML runtimes
  • Identify how AutoML facilitates model/feature selection.
  • Identify the advantages AutoML brings to the model development process
  • Identify the benefits of creating feature store tables at the account level in Unity Catalog in Databricks vs at the workspace level
  • Create a feature store table in Unity Catalog
  • Write data to a feature store table
  • Train a model with features from a feature store table.
  • Score a model using features from a feature store table.
  • Describe the differences between online and offline feature tables
  • Identify the best run using the MLflow Client API.
  • Manually log metrics, artifacts, and models in an MLflow Run.
  • Identify information available in the MLflow UI
  • Register a model using the MLflow Client API in the Unity Catalog registry
  • Identify benefits of registering models in the Unity Catalog registry over the workspace registry
  • Identify scenarios where promoting code is preferred over promoting models and vice versa
  • Set or remove a tag for a model
  • Promote a challenger model to a champion model using aliases

Lessons

  1. ML runtimes and AutoML
  2. Tracking experiments with MLflow
  3. Models in Unity Catalog: register, tag and alias
  4. Feature tables in Unity Catalog
  5. Training and scoring with feature tables; online vs offline
  6. MLOps strategy: environments, deploy code vs deploy models
Section 2 · Data Processing19%

What the exam asks

  • Compute summary statistics on a Spark DataFrame using .summary() or dbutils data summaries
  • Remove outliers from a Spark DataFrame based on standard deviation or IQR
  • Create visualizations for categorical or continuous features
  • Compare two categorical or two continuous features using the appropriate method
  • Compare and contrast imputing missing values with the mean or median or mode value
  • Impute missing values with the mode, mean, or median value
  • Use one-hot encoding for categorical features
  • Identify and explain the model types or data sets for which one-hot encoding is or is not appropriate.
  • Identify scenarios where log scale transformation is appropriate

Lessons

  1. Summary Statistics and Outlier Removal
  2. Visualizing, Comparing and Log-Transforming Features
  3. Imputing Missing Values and One-Hot Encoding
Section 3 · Model Development31%

What the exam asks

  • Use ML foundations to select the appropriate algorithm for a given model scenario
  • Identify methods to mitigate data imbalance in training data
  • Compare estimators and transformers
  • Develop a training pipeline
  • Use Hyperopt's fmin operation to tune a model's hyperparameters
  • Perform random or grid search or Bayesian search as a method for tuning hyperparameters.
  • Parallelize single node models for hyperparameter tuning
  • Describe the benefits and downsides of using cross-validation over a train-validation split.
  • Perform cross-validation as a part of model fitting.
  • Identify the number of models being trained in conjunction with a grid-search and cross-validation process.
  • Use common classification metrics: F1, Log Loss, ROC/AUC, etc
  • Use common regression metrics: RMSE, MAE, R-squared, etc.
  • Choose the most appropriate metric for a given scenario objective
  • Identify the need to exponentiate log-transformed variables before calculating evaluation metrics or interpreting predictions
  • Assess the impact of model complexity and the bias variance tradeoff on model performance

Lessons

  1. Choosing an Algorithm and Handling Imbalanced Classes
  2. Estimators, Transformers and Training Pipelines
  3. Hyperparameter Tuning: Grid, Random and Bayesian Search with Hyperopt and Optuna
  4. Cross-Validation, Counting Models and the Bias-Variance Trade-off
  5. Evaluation Metrics and Log-Transformed Targets
Section 4 · Model Deployment12%

What the exam asks

  • Identify the differences and advantages of model serving approaches: batch, realtime, and streaming
  • Deploy a custom model to a model endpoint
  • Use pandas to perform batch inference
  • Identify how streaming inference is performed with Delta Live Tables
  • Deploy and query a model for realtime inference
  • Split data between endpoints for realtime interference

Lessons

  1. Batch, Streaming and Real-Time Inference; Batch Scoring with pandas
  2. Streaming Inference with Lakeflow Spark Declarative Pipelines
  3. Real-Time Inference with Model Serving: Deploy, Query and Split Traffic
Study plan

14-Day Standard Plan

The default path for data scientists and ML engineers who know Python and basic machine learning but haven't used every Databricks ML feature the exam covers. It covers all 17 lessons and all six labs in section order, with a section drill at the end of each section and short daily flashcard reviews. Sections 1 (38%) and 3 (31%) get four days each, because together they carry about 33 of the 48 scored questions. Day 12 closes the content with the deployment lab, Mock Exam 1 is on day 13 and Mock Exam 2 on day 14, and you sit the exam on day 15. Run every drill with Keep the mock-exam questions unseen ticked, so both mocks are new when you take them. About 1.5–2 hours a day.

  1. Day 1Kick-off: the exam, ML runtimes, AutoML and MLflow tracking
  2. Day 2Models in Unity Catalog, then the MLflow lab
  3. Day 3Feature tables: create, write, train and score
  4. Day 4MLOps strategy, the feature table lab and the m1 drill
  5. Day 5Profiling, outliers, charts and log transforms
  6. Day 6Imputation and one-hot encoding, the m2 lab and drill
  7. Day 7Algorithms, class imbalance, estimators and pipelines
  8. … and 7 more days, with mock exams at the end

Read the free study guide for this exam

Try it

Sample questions

Written for practice in the style of the real exam: never questions from the real exam.

Sample question 1

A training set uses two feature tables, customer_features and merchant_features. Both contain a feature named avg_amount_90d, one computed per customer and one per merchant, and the model needs both. How should the data scientist build the training set?

  1. ASet rename_outputs in one FeatureLookup, for example {"avg_amount_90d": "merchant_avg_amount_90d"}
  2. BPass exclude_columns=["avg_amount_90d"] to create_training_set, which keeps both copies under generated names
  3. CSet feature_names=None in both lookups, so the client adds table-name prefixes to features whose names clash
  4. DRename one of the columns in pandas after load_df() and train on the renamed DataFrame
Show the answer

A. Each FeatureLookup can rename its outputs with a dict of feature name to new name (default_values must then use the new names). The renamed feature is stored in the feature spec that fe.log_model packages, so batch scoring and serving produce the same column names the model was trained on.

Sample question 2

A CrossValidator is configured without setting numFolds, and its ParamGridBuilder grid has 2 values of regParam and 5 values of elasticNetParam. How many models are trained during cross-validation, before the final refit?

  1. A50
  2. B30
  3. C10
  4. D21
Show the answer

B. Spark ML's CrossValidator uses numFolds=3 by default, while scikit-learn's search classes use 5 folds when cv=None. The count is combinations x folds = (2 x 5) x 3 = 30, and the refit of the best combination adds one more.

Questions

About this exam

What is the passing score?

Databricks doesn't publish it. This course's readiness target is ≥ 80% (39 of 48) on two mocks on the first attempt, with ≥ 70% in every section.

How many questions come from each section?

Databricks publishes only the weights. On 48 scored items they come to about 18 (Databricks Machine Learning, 38%), 9 (Data Processing, 19%), 15 (Model Development, 31%) and 6 (Model Deployment, 12%). Exam forms vary slightly.

Why does the exam page call Section 2 "ML Workflows"?

The exam page's weighting list labels it ML Workflows, while the exam guide's outline calls it Data Processing. Both are 19%, and the objectives are the guide's: summary statistics, outliers, visualization, imputation, one-hot encoding and log transforms.

Are there unscored questions?

Possibly. Up to 10 unscored items may be added. They look like any other item and don't count, and the extra time is already in the 90 minutes. Treat every item as scored.

How do multiple-selection questions work?

The stem says how many to choose, such as "(Choose two.)". Databricks doesn't say whether partial credit is given, so assume an item counts only when every option you pick is right. About one question in seven in this course is multiple-select, like the exam's.

What language is the code in?

The exam page says machine learning code is in Python; non-ML code may be in SQL. Expect scikit-learn, Spark ML (pyspark.ml), pandas, MLflow and the Feature Engineering client.

Hyperopt is deprecated. Is it still on the exam?

Yes. The guide lists "Use Hyperopt's fmin operation", so learn fmin, tpe.suggest, the hp.* expressions and SparkTrials. In practice Hyperopt isn't in Databricks Runtime ML after 16.4 LTS ML, and Databricks recommends Optuna or Ray Tune.

The exam says Delta Live Tables. Which name should I expect?

Either. The product is now Lakeflow Spark Declarative Pipelines, but the guide's objective and sample question say Delta Live Tables. The pattern is the same: the model as a spark_udf inside a streaming table.

Which runtime should I practise on, now that AutoML isn't built into 18 LTS ML?

Most labs run on Free Edition serverless compute. Classic AutoML needs a paid workspace with a 17.3 LTS ML cluster in Dedicated access mode; 17.3 LTS is supported until October 2028. On the exam, answer as the guide expects: the ML runtime includes AutoML.

Can I do every lab on Databricks Free Edition?

Almost. The optional online feature store and AutoML steps need a paid workspace, and Free Edition limits serving endpoints (CPU only, a limited number active at once) and pipelines (one active per type). Delete endpoints when you finish a lab.

Ready to start?

See where you stand in 8 minutes

Ten questions from this course, weighted like the real exam, with your score by section and a rough chance of passing.

Take the free check