ICDL Data Analytics (ICDL) flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

Descriptive analytics

Summarises historical data to show what happened, using totals, averages, rates and trends. It is the simplest of the four analytics types and answers no causal question.

Predictive analytics

Uses statistical models and historical patterns to estimate the likelihood of future outcomes, such as forecasting demand or scoring customers by churn risk.

Nominal data

Categorical data with no inherent order, such as country, product colour or department. Only frequency counts and the mode are meaningful summaries.

Ratio data

Numeric data with equal intervals and a true zero, such as revenue, weight or duration, so all arithmetic including ratios and percentages is valid.

Population versus sample

A population is every unit of interest; a sample is a subset measured instead. Parameters describe populations and statistics describe samples.

Median

The middle value of an ordered dataset, or the average of the two middle values when the count is even. It is resistant to outliers.

Correlation does not imply causation

A high correlation coefficient shows association only. Confounding variables, reverse causation or coincidence may explain the relationship, so experiments or further evidence are needed.

KPI

Key performance indicator: a metric tied to a strategic objective with a defined target, an accountable owner and a regular reporting cadence.

Data quality dimensions

Accuracy, completeness, consistency, timeliness, validity and uniqueness. Each is assessed separately because a dataset can score well on one and poorly on another.

ELT versus ETL

ETL transforms data before loading it into the warehouse; ELT loads raw data first and applies transformations inside the warehouse using its own compute power.

Third normal form

A table is in 3NF when it is already in 2NF and contains no transitive dependencies, so every non-key field depends only on the key, not on other non-key fields.

IQR outlier rule

Values below Q1 minus 1.5 times the interquartile range, or above Q3 plus 1.5 times the IQR, are flagged as potential outliers for review.

Left join

A join returning all rows from the left table and only matching rows from the right table; unmatched right-side fields are filled with null values.

Mean imputation

Replacing missing values with the variable's average. It preserves sample size but shrinks variance and can weaken or distort correlations with other variables.

Referential integrity

The rule that every foreign key value in a child table must match an existing primary key in a parent table, which prevents orphan records.

Primary key

A column, or combination of columns, that uniquely identifies every row in a table and cannot contain nulls or duplicate values.

COUNT versus COUNTA

COUNT counts cells containing numbers only, whereas COUNTA counts every non-empty cell, including text, dates and logical values.

SUMIFS function

Adds values in a sum range where one or more criteria ranges match their criteria, using operators and wildcards such as ">=1000" or "North*".

VLOOKUP limitation

VLOOKUP searches only to the right of the lookup column and defaults to approximate match, so an exact match needs FALSE and left lookups fail.

PivotTable

An interactive summary that groups rows by category fields and aggregates a value field using sum, count or average, requiring a refresh after source data changes.

Simpson's paradox

An aggregate trend that reverses when the data is split into subgroups, typically because subgroup sizes are unbalanced across the compared categories.

R-squared

The proportion of variance in the dependent variable explained by a regression model, ranging from 0 (explains nothing) to 1 (perfect fit).

P-value

The probability of observing data at least as extreme as the sample if the null hypothesis were true. It is not the probability that the null is true.

Moving average

The average of the most recent n periods, recalculated each period as the window advances, used to smooth short-term noise and expose underlying trend.

Line chart

Best for continuous time series: time runs along the horizontal axis and the slope of the line communicates the rate of change between periods.

Scatter plot

Plots one numeric variable against another to reveal the direction and strength of a relationship, plus clusters and outlying observations.

Truncated axis

Starting a value axis above zero, which exaggerates differences between bars or columns. Bar and column charts should always include zero.

Sequential colour scale

A single hue varying from light to dark that encodes ordered low-to-high values; diverging scales are used instead when data has a meaningful midpoint.

Data-ink ratio

Tufte's principle that the share of ink representing actual data should be maximised, so gridlines, borders and decoration without informational value are removed.

Five-second rule

A dashboard's key message and headline metrics must be graspable within about five seconds, so the most important indicator is placed top-left.

Alt text

A short written description of a chart or image, read aloud by screen readers, conveying the key insight rather than just naming the chart type.

Histogram versus bar chart

A histogram bins continuous data into intervals and shows adjacent bars with no gaps; a bar chart compares discrete categories and keeps gaps between bars.