BigQuery ML ARIMA_PLUS for quick monthly sales forecasts

PMLE By islever Editorial Last updated

Short answer

C Build a BigQuery ML ARIMA_PLUS model with time_series_id_col set to product ID to forecast monthly sales straight from the existing BigQuery data.

Question

You work for an online retailer. Your company has a few thousand short lifecycle products. Your company has five years of sales data stored in BigQuery. You have been asked to build a model that will make monthly sales predictions for each product. You want to use a solution that can be implemented quickly with minimal effort. What should you do?
  1. Use Prophet on Vertex AI Training to build a custom model.
  2. Use Vertex AI Forecast to build a NN-based model.
  3. Use BigQuery ML to build a statistical ARIMA_PLUS model. Correct Answer
  4. Use TensorFlow on Vertex AI Training to build a custom model.

Key Takeaways

It tests recognising when a managed SQL-based ARIMA_PLUS forecast in BigQuery ML meets the 'quick, minimal effort' requirement better than any Vertex AI custom training option.

This PMLE question tests how to choose a fast, low-effort forecasting path when sales data already sits in BigQuery. The answer is (C), a BigQuery ML ARIMA_PLUS model, rather than a custom Vertex AI training job.

Common mistake: Choosing B, Vertex AI Forecast with an NN-based model, because it sounds automated and scalable; a neural model still needs feature engineering, a long training run and deployment, so it is not the minimal-effort path.

Expert Analysis

The boundary in question The trap here is the line between a SQL-native, fully managed statistical forecast and a custom training job on Vertex AI. "Implemented quickly with minimal effort" is the deciding constraint, and it tells you to stay on the BigQuery side of that line. Correct answer: C.

Why C stays on the quick side Your five years of sales history already live in BigQuery, so nothing needs exporting or moving. A single CREATE MODEL statement with model_type = 'ARIMA_PLUS' and time_series_id_col = product_id trains an independent series for each of your few thousand products in one go, with no clusters, containers or hyperparameter tuning. ARIMA_PLUS automatically detects trend and seasonality, can model holiday effects, and returns monthly forecasts through a simple ML.FORECAST with your chosen horizon — exactly the shape of this problem.

Why A, B and D cross it A (Prophet) and D (TensorFlow) both require Vertex AI Training jobs: writing and packaging code, provisioning infrastructure, then adding a separate serving step. B sounds managed, but the option explicitly specifies an NN-based model, which means a long-running neural training cycle plus feature engineering and deployment — the opposite of minimal effort, and far slower to first prediction than a statistical model.

The trap "A few thousand products" tempts you toward Vertex AI's scale, but ARIMA_PLUS already handles thousands of series natively, so scale does not justify the extra effort.

Official Reference

https://cloud.google.com/bigquery-ml/docs/reference/standard-sql/bigqueryml-syntax-create

This link opens the official vendor documentation in a new tab.

Exam Strategy

On PMLE questions, treat phrases like 'quickly', 'minimal effort' and 'data already in X' as selection criteria that rank managed SQL-only options above custom training jobs. First check whether the data already resides in BigQuery — if so, shortlist BigQuery ML before considering Vertex AI Training or Vertex AI Forecast. Then confirm the option matches the workload shape (time series, per-product ID column, monthly horizon) so you do not pick a managed service whose specified model type reintroduces heavy effort.

Frequently Asked Questions

Why does BigQuery ML ARIMA_PLUS fit 'minimal effort' better than Vertex AI Forecast?

Because the data already lives in BigQuery: a single CREATE MODEL statement trains and stores the model with no data export, container packaging or separate serving stack. ARIMA_PLUS also auto-detects trend, seasonality and holiday effects, so no feature engineering or neural architecture search is needed.

Can ARIMA_PLUS forecast thousands of separate products at once?

Yes. Setting time_series_id_col to the product identifier trains one independent ARIMA_PLUS series per product inside the same CREATE MODEL statement, and ML.FORECAST then returns a forecast for every series with the requested horizon.

When would Vertex AI Forecast or Prophet be the better choice?

Move to them when you need deep learning accuracy across many correlated series, exogenous regressors, custom training controls, or holiday/regressor tuning beyond ARIMA_PLUS. That extra capability costs setup time, training time and a separate deployment step.

More PMLE FAQ →

Community Votes

C
100%

100% of anonymous learners picked answer C. Votes are pick records left by other test-takers — they are not the verified answer.

Discussion 6 contributions

Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.

From public community threads 6

  1. pikachu007 👍 5 Answered C
    Ease of Use: BigQuery ML integrates seamlessly with BigQuery, allowing you to create and train models directly within SQL queries, eliminating the need for separate environments or coding. Statistical ARIMA_PLUS Strengths: This model is well-suited for time series forecasting, automatically handling seasonality, trends, and holidays, making it appropriate for monthly sales predictions. Minimal Effort: BigQuery ML handles model training and tuning, reducing the need for manual configuration or hyperparameter tuning. Fast Implementation: Model creation and training can be done in a few lines of SQL, enabling rapid deployment.
  2. fitri001 👍 3 Answered C
    Quick Implementation: BigQuery ML simplifies the process. You can train and deploy the model directly within BigQuery, eliminating the need for complex model deployment or data movement. Minimal Effort: ARIMA_PLUS is a pre-built statistical model available in BigQuery ML. You don't need to write custom code for a complex neural network (NN) model like in option B or D. Time Series Data: ARIMA models are well-suited for time series forecasting, which is ideal for your monthly sales prediction task.
  3. fitri001 👍 2
    why not others? A. Prophet on Vertex AI Training: While Prophet is a good choice for time series forecasting with holidays and seasonality, using Vertex AI Training requires additional setup and potentially custom code compared to the readily available ARIMA_PLUS model within BigQuery ML. B. Vertex AI Forecast with NN-based Model: Building a custom NN-based model using Vertex AI Forecast offers flexibility but requires more effort and expertise in model development and potentially hyperparameter tuning. This might not be ideal for a quick implementation. D. TensorFlow on Vertex AI Training: Similar to option B, using TensorFlow for a custom model offers flexibility but requires significant coding and expertise, making it less suitable for a quick and low-effort approach.
  4. taylz876 👍 1 Answered C
    Why C is right The sales data is already in BigQuery, and you need a solution that can be implemented quickly with minimal effort. BigQuery ML trains the model with a few lines of SQL, with no data movement and no infrastructure. A few thousand products means a few thousand separate time series. Setting time_series_id_col to the product ID lets a single CREATE MODEL statement train an independent forecast for every product in parallel. Monthly predictions from five years of history is a classic ARIMA_PLUS job. Why the others are wrong • A and D: Prophet or TensorFlow on Vertex AI Training means writing and running custom model code for thousands of products, which is the opposite of quick and minimal effort. • B: Vertex AI Forecast with a neural network model needs the data exported, a dataset set up and a long training job. It is also heavier than a SQL statement for this simple forecasting task. • "ARIMA_PLUS (univariate) → forecasts a column using only its own history" • time_series_id_col: "ID columns; each combination = a separate time series. One CREATE MODEL trains an independent forecast per country/product pair, run in parallel (scales to huge numbers of series)." • Exam Tips: "Forecast using only past values → ARIMA_PLUS" and "Many products/regions in one statement → time_series_id_col." • Cheat Sheet: "Many series at once → time_series_id_col."
  5. b1a8fae 👍 1 Answered C
    Given amount of data (few thousand short-cycled products) and frequency of predictions (monthly) C is the way to go.
Show 1 more comment
  1. pinimichele01 👍 1 Answered C
    data on bigquery + minimal effort -> C

Reader comments & corrections

No reader corrections yet — spotted an error or have a note? Share it below.

Log in to comment, report an error, or add a note about this question.

Practice All PMLE Questions

Access 65 questions with complete answers and detailed explanations.

View Full PMLE Practice Test →

← Back to PMLE Study Guide