Conversational Analytics Code Interpreter

The Code Interpreter in Conversational Analytics translates your natural language questions into Python code and executes that code to provide advanced analysis and visualizations. The Code Interpreter is available for Conversational Analytics in Looker Studio and Looker (original).

In contrast to standard SQL-powered BI experiences, the Code Interpreter supports a wide variety of data analytics—from basic computations and charting to more advanced tasks like time series forecasting. The Code Interpreter enhances Conversational Analytics by enabling users to perform these types of advanced analysis, which otherwise would typically require specialized knowledge of advanced coding or statistical methods.

Learn how and when Gemini for Google Cloud uses your data. As an early-stage technology, Gemini for Google Cloud products can generate output that seems plausible but is factually incorrect. We recommend that you validate all output from Gemini for Google Cloud products before you use it. For more information, see Gemini for Google Cloud and responsible AI.

Before you begin

To use the Code Interpreter, you must meet the requirements for using Conversational Analytics in Looker Studio or in Looker (original).

To use the Code Interpreter with Looker, your instance must be a Looker (original) instance on Looker version 25.8 or later.

Enable the Code Interpreter

To enable the Code Interpreter for your conversations, follow these steps:

  1. In the left navigation panel within Conversational Analytics, click the Advanced analytics toggle to enable the Code Interpreter.
  2. With the Code Interpreter enabled, you can use Conversational Analytics as usual to start conversations and ask questions of your data. The Code Interpreter uses the engine that powers Gemini chat to translate your queries into Python code and execute that code.

Known limitations

  • The Code Interpreter uses Python to solve problems. Since Python is more flexible than structured query languages, Code Interpreter responses might have more variability than responses from the core Conversational Analytics experience.
  • For Looker data, Conversational Analytics can return a maximum of 5,000 rows per query.
  • The Code Interpreter supports these Python libraries. To request support for additional Python libraries, send an email to conversational-analytics-feedback@google.com.
  • The following visualization chart types are not supported in Code Interpreter responses:
    • Maps

For information about additional limitations, see the documentation on known limitations in Conversational Analytics.

Supported Python libraries

Show supported Python libraries

The Code Interpreter supports the following Python libraries:

  • altair
  • attrs
  • chess
  • contourpy
  • cycler
  • entrypoints
  • fonttools
  • fpdf
  • geopandas
  • imageio
  • jinja2
  • joblib
  • jsonschema
  • jsonschema-specifications
  • kiwisolver
  • lxml
  • markupsafe
  • matplotlib
  • mpmath
  • numexpr
  • numpy
  • opencv-python
  • openpyxl
  • packaging
  • pandas
  • patsy
  • pdfminer-six
  • pillow
  • plotly
  • protobuf
  • pylatex
  • pyparsing
  • PyPDF2
  • python-dateutil
  • python-docx
  • python-pptx
  • pytz
  • referencing
  • reportlab
  • rpds-py
  • scikit-image
  • scikit-learn
  • scipy
  • seaborn
  • six
  • statsmodels
  • striprtf
  • sympy
  • tabulate
  • tensorflow
  • threadpoolctl
  • toolz
  • torch
  • tzdata
  • xlrd

Suggested questions

When you enable the Code Interpreter, Python's advanced analytic capabilities enable Conversational Analytics to answer a wider range of questions, in addition to the standard types of supported questions. For example:

  • Can you explain the key drivers for sales based on my data?
  • What is the lifetime value for each of my customer segments, taking into account the average purchase frequency and the average order value?
  • How do sales this year compare to sales last year?
  • Identify outliers in my sales data to help identify products or regions that are performing particularly well or particularly poorly.
  • Perform a cohort analysis to understand customer retention.
  • Are my highest margin products also the most popular products? Use this answer to provide a suggestion on how to optimize my product mix.
  • What is the compound annual growth rate (CAGR) for sales by product category for the last 3 years?
  • Show the CAGR as a bar chart with product category on the x-axis and CAGR on the y-axis.