Drag and Drop Question You plan to explore demographic data for home ownership in various cities. The data is in a CSV file with the following format: age,city,income,home_owner 21,Chicago,50000,0 35,

Question

Exhibit

Answer Area

Explanation

The question tests the ability to correctly apply Azure ML Run object logging methods for different data types: scalar values, plots, and structured dictionaries.

Approach. The core task is to log three distinct types of information using the appropriate methods of the run object in Azure Machine Learning:

1. Logging the number of observations (a single numerical value): The code calculates row_count = len(data). To log a single key-value metric, the run.log() method is used. The blank for this requirement is run. Segment ("observations", row_count). Therefore, 'log' should be dragged here.
2. Logging a box plot (a Matplotlib figure): The code generates a Matplotlib figure named fig. To log an image or a plot, the run.log_image() method is specifically designed for this purpose. The blank for this requirement is run. Segment (name = "income_by_home_owner", plot = fig). Therefore, 'log_image' should be dragged here.
3. Logging a dictionary of city names and average income: The code creates mean_inc_dict, which is a dictionary mapping city names to average incomes. To log structured data like a dictionary or tabular data, the run.log_table() method is the most appropriate. It can interpret a dictionary as key-value pairs suitable for a table. The blank for this requirement is run. Segment (name = "mean_income_by_city", value = mean_inc_dict). Therefore, 'log_table' should be dragged here.

Common mistakes.

• common_mistake. Common mistakes arise from misunderstanding the purpose of each logging method:
• Using 'log_list': log_list is used to log a list of numeric values or strings under a single metric name. It's not suitable for a single scalar value, a plot, or a key-value dictionary that represents structured data.
• Using 'log_row': log_row is used to log a single row of a table. While a dictionary could be interpreted as a row, log_table is generally more versatile and idiomatic for logging a complete dictionary where keys often represent columns or distinct items.
• Using 'log' for a plot or dictionary: The log method is strictly for logging single key-value metrics (scalars). It cannot handle Matplotlib figure objects or complex dictionary structures directly for visualization or structured storage.
• Interchanging 'log_table' with 'log_list' or 'log_row' for the dictionary: The mean_inc_dict is structured data, and log_table is the most appropriate for representing such data in a structured, queryable format within the experiment run.

Concept tested. Azure Machine Learning experiment tracking and logging, specifically using the azureml.core.Run object's methods to log different data types (scalars, images/plots, and structured data like dictionaries) to track experiment metrics and artifacts.

No community discussion yet for this question.