ICDL Data Analytics (ICDL) key terms

The practice of examining raw information to find patterns, trends, and insights that support conclusions or decisions. It combines statistical methods, software tools, and business context rather than simple counting.

The branch of analytics that summarises what has already happened, using totals, averages, trends, and dashboards. It answers questions about past performance and forms the base for deeper investigation.

Analysis aimed at explaining why something happened by comparing segments, checking correlations, and drilling into detail. It moves beyond summary figures to identify contributing causes and anomalies.

Techniques that use historical data and statistical models to estimate future outcomes such as demand, churn, or risk. Results are probabilities or forecasts, not certainties, and depend on data quality.

Analysis that recommends actions, for example by optimising schedules or simulating scenarios. It builds on predictions and rules to suggest the best option under stated constraints.

Information expressed as numbers that can be measured and calculated, such as revenue, temperature, or headcount. It supports arithmetic operations and statistical summary.

Non-numeric information such as interview comments, categories, or descriptions. It captures meaning and context and is usually analysed by grouping, coding, or counting occurrences rather than averaging.

Numeric values restricted to whole counts, such as number of orders or employees. Values between the counted points are not meaningful, which affects which charts and averages are appropriate.

Numeric values that can take any value within a range, such as time, weight, or temperature. Measurements are limited only by the precision of the recording instrument.

Values that place records into named groups, such as region, product type, or status. Categories may be ordered or unordered and are summarised with counts and proportions rather than averages.

Information collected first-hand for the specific question at hand, for example through a survey or sensor. The collector controls definitions and timing but usually bears the full collection cost.

Information gathered by someone else for another purpose and reused, such as published statistics or internal sales records. It is quicker and cheaper but may not match the current question exactly.

The origin of a dataset, such as a database, spreadsheet, web service, or manual entry form. Recording the source supports verification, refresh, and assessment of reliability.

The degree to which data is accurate, complete, consistent, and fit for its intended use. Poor quality produces misleading charts and unreliable conclusions regardless of the analysis technique used.

The process of detecting and correcting errors, duplicates, inconsistent formats, and missing values before analysis. It is often the most time-consuming stage and directly affects result credibility.

An observation that lies far from the rest of the data. It may signal a recording error or a genuine unusual event, so it should be investigated before being removed or retained.

A blank or absent entry where data was expected. Analysts must decide whether to exclude, impute, or flag it, because default handling can bias totals and averages.

A characteristic or field recorded for each record in a dataset, such as age, price, or region. Variables are classified by type, which determines the valid summaries and charts.

One row or entry in a dataset describing a single entity, such as one customer or one transaction. Each record holds values for the same set of variables.

An organised collection of related records and variables stored together for analysis. It may live in a spreadsheet, database table, or file and should be documented with its source and definitions.

The arithmetic average, found by adding all values and dividing by the count. It uses every observation and is sensitive to extreme values.

The middle value when observations are sorted. Half the values lie above and half below, making it resistant to outliers and useful for skewed distributions.

The value or category that occurs most often in a dataset. It is the only average suitable for unordered categorical data such as product type or region.

A measure of how widely values spread around the mean. A small value indicates clustering near the average; a large value indicates dispersion and less predictable individual values.

The average of the squared differences between each value and the mean. It quantifies spread but is expressed in squared units, so its square root gives the standard deviation.

A summary showing how many observations fall into each category or interval. It reveals the shape of the data, including common values and gaps, and underpins histograms and bar charts.

A statistical measure of the strength and direction of a linear relationship between two numeric variables, usually ranging from -1 to +1. It does not by itself prove that one variable causes the other.

Examining data over time to identify direction, seasonality, or turning points. It often uses line charts and moving averages to separate underlying movement from short-term noise.

A series of averages calculated over a fixed number of consecutive periods, recalculated as the window slides forward. It smooths short-term fluctuations to make the underlying trend easier to see.

A chart using rectangular bars whose lengths represent values for categories. It compares discrete groups well, and horizontal orientation suits long category labels.

A chart with vertical bars showing values for categories, often used for comparisons across groups or across a few time periods. Grouped or stacked variants show sub-categories.

A chart that joins data points with lines, typically with time on the horizontal axis. It is best for showing continuous change and trends rather than comparing unrelated categories.

A circular chart divided into slices whose angles represent parts of a whole. It works only when categories sum to 100 percent and there are few slices; otherwise comparison becomes difficult.

A chart plotting one numeric variable against another to reveal relationships, clusters, and outliers. Adding a trend line helps show the direction of association.

A chart grouping continuous numeric data into intervals and showing the count in each. Adjacent bars touch because the horizontal axis is a continuous scale, unlike a bar chart.

A bar, column, or area chart where sub-category values are placed on top of each other to show both the total and its composition. It suits part-to-whole comparison across several groups.

A line fitted through plotted data points to summarise the general direction of a relationship. It aids interpretation but does not guarantee that future points will follow it.

Rules applied to input cells or fields that restrict what can be entered, such as allowed lists or number ranges. It prevents many errors at the point of entry rather than after analysis.

Arranging records in a defined order, such as ascending date or descending value. It supports ranking, spotting duplicates, and preparing data for reports and lookups.

Hiding records that do not meet specified criteria so only relevant rows are displayed or analysed. It narrows focus without deleting the underlying data.

An interactive summary tool that groups and aggregates records by chosen row, column, and value fields. It allows rapid cross-tabulation and can be reconfigured without rewriting formulas.

Combining many detailed records into summary figures such as sums, counts, or averages by group. It reduces volume for reporting but hides individual-level detail.

A formula that finds a value in one table and returns a related value from another, such as matching a product code to its price. It links datasets without manual copying.

A store of data organised into tables linked by shared key fields. Relationships let related information be combined accurately while avoiding unnecessary duplication.

A field or combination of fields that uniquely identifies each record in a table. It prevents duplicate rows and provides a reliable link to related tables.

A single screen presenting key indicators and charts, often refreshed automatically from underlying data. Good dashboards focus on a few decision-relevant measures rather than every available number.

A key performance indicator: a measure chosen because it directly reflects progress toward an important objective. Effective KPIs are specific, measurable, and tied to a target.

A reference value or standard against which performance is compared, such as a target, prior period, or competitor figure. Without a benchmark, a number alone rarely indicates success or failure.

Comparing actual results with a budget, forecast, or prior period to explain the size and cause of differences. It directs attention to the gaps that matter most.

Changing input assumptions to see how outputs respond, often using scenarios or data tables. It helps decision makers understand sensitivity and possible ranges of outcomes.

Presenting data graphically so patterns and comparisons are understood quickly. Effective visuals match the chart type to the data and remove decoration that does not aid interpretation.

Combining accurate visuals with a clear narrative that explains what the data shows and why it matters. It guides an audience from evidence to a recommended action.

The tendency to notice and favour evidence that supports existing beliefs while discounting contrary data. Analysts guard against it by testing alternative explanations and reviewing results critically.

A proportion expressed as a number out of one hundred. It allows fair comparison between groups of different sizes, but small bases can make percentages misleading.