Google Data Analytics key terms

The interacting collection of people, tools, processes, and technologies that produce, store, manage, analyze, and share data. Understanding the ecosystem helps an analyst see where data comes from and who depends on it.

The process of collecting, cleaning, transforming, and interpreting information to answer a question or support a decision. It focuses on the reasoning and methods used rather than on the tools alone.

Information expressed as numbers that can be counted, measured, or calculated, such as revenue, age, or number of visits. It supports statistical summaries, averages, and comparisons across groups or time periods.

Descriptive, non-numeric information that captures qualities, opinions, or categories, such as interview comments or product colors. It is often coded or grouped so patterns can be summarized and compared.

Data with no fixed, table-like format, such as emails, videos, audio, or free-text notes. It usually requires extra preparation, like text parsing or tagging, before it can be analyzed alongside structured sources.

Choosing a course of action based on facts gathered and analyzed rather than on intuition or seniority alone. The process normally moves from asking a question, to gathering and cleaning data, to analysis, to action.

The set of principles governing how data should be collected, used, and shared fairly, including honesty, privacy, and avoiding harm. Ethical choices shape which data is gathered and how findings are presented to others.

The process of breaking a vague problem into specific, ordered pieces and defining what information each piece needs. It keeps an analysis focused on the original question instead of drifting into unrelated exploration.

Questions framed to be Specific, Measurable, Actionable, Relevant, and Time-bound so that an answer can actually guide a decision. Rewriting a vague request into a SMART question is a common first step in an analysis.

A person or group who invests in a project, uses its results, or is affected by the decisions it informs. Clarifying stakeholder expectations early prevents analysis that answers a question nobody asked.

The specific question or need an analysis must satisfy, stated in clear and neutral terms. It defines scope, success criteria, and what data is relevant, and it should be confirmed with stakeholders before work begins.

A method of comparing the current state of something with the desired state to reveal what is missing or underperforming. The identified gaps then become targets for further investigation and action.

Data collected first-hand by the analyst or organization for the specific question at hand, for example through surveys, interviews, or sensors. It is tailored and current but typically costs more time and effort to gather.

Data that was originally collected by someone else for a different purpose, such as public reports, industry databases, or internal archives. It is usually faster and cheaper to obtain but may be outdated or imperfectly suited.

Data that describes other data, telling you how, when, where, and by whom a dataset was created and what its fields mean. It helps analysts judge whether a source is trustworthy and how to interpret it correctly.

A reference document listing each field in a dataset with its meaning, format, allowed values, and any related table. It helps analysts and colleagues use the data consistently and avoid misreading column names.

A store of data organized into tables that are linked by shared key fields, allowing information to be combined without duplication. Queries retrieve and combine rows from those tables to answer questions.

An operation that combines rows from two or more tables using a shared field, such as a customer ID, to produce one wider result. Choosing the correct join type determines which records are kept or dropped.

The policies, roles, and standards that control how data is stored, accessed, and used across an organization. Good governance protects data quality and ensures people can find and trust the data they need.

The practice of handling personal information responsibly, including limiting collection, controlling access, and respecting consent and regulation. Analysts must protect identifiers and share only aggregated or de-identified results where appropriate.

Structured Query Language, the standard language for retrieving, filtering, joining, and summarizing data held in relational databases. Analysts use queries to pull exactly the records and fields an analysis requires.

A distortion that occurs when the group of records or people studied is not representative of the population being described. Conclusions then reflect the sample's quirks rather than the real-world pattern of interest.

Data containing errors, duplicates, missing entries, inconsistent formats, or irrelevant records that would distort results if used unchanged. It must be identified and corrected before reliable analysis can proceed.

The process of finding and fixing problems in a dataset, such as removing duplicates, standardizing formats, correcting typos, and handling blanks. The goal is data that is accurate, complete, and consistent enough to analyze.

Checking that data meets defined rules for format, range, type, or completeness before it is used or stored. Validation catches problems at entry or import, reducing the amount of corrective cleaning needed later.

A value that lies far outside the range of most other observations in a dataset. Outliers may signal a data entry error, a measurement problem, or a genuinely unusual case, so each one needs individual judgment.

Changing the structure, scale, or format of data so it can be compared or combined, for example converting units, splitting columns, or standardizing date formats. Transformations should be documented so results can be reproduced.

Summarizing many individual records into a smaller set of totals, averages, counts, or other grouped figures. It makes large datasets readable and is usually done by grouping records by a category such as month or region.

A spreadsheet tool that groups and summarizes data by dragging fields into rows, columns, and values, producing a compact cross-tabulation. It allows analysts to compare totals quickly and to slice data in several ways.

A measurable relationship in which two variables tend to move together, either in the same direction or in opposite directions. Correlation alone does not show that one variable causes the other.

Presenting data as charts, graphs, maps, or other visuals so patterns and comparisons are easier to see than in a raw table. Good visuals highlight the key finding and reduce the effort a viewer spends interpreting it.

A single screen that combines several charts and key metrics, often updated automatically, so stakeholders can monitor performance at a glance. It is designed for quick ongoing tracking rather than deep narrative explanation.

Combining data, visuals, and narrative to explain what was found, why it matters, and what should happen next. A clear story connects the evidence to a recommended action for a specific audience.

Visual properties such as color, size, shape, and position that the eye registers almost instantly, before conscious reading. Using them deliberately directs attention to the most important part of a chart.

A two-dimensional, labeled table structure in the pandas library, with named columns and an index, similar to a spreadsheet held in memory. It supports filtering, grouping, joining, and summarizing data with short commands.

An interactive document that mixes runnable code cells, their outputs, and written notes in one file. It is widely used in Python analysis because it records both the steps taken and the reasoning behind them.

A structure for running an analytics project in four stages: Plan, Analyze, Construct, and Execute. It keeps the business question, the data work, and the final deliverable aligned from start to finish.

A curated collection of completed analytics projects that demonstrates practical skills, methods, and communication to employers or clients. Strong cases state the question, show the data work, and explain the resulting recommendation.