Measures and metrics are useful business tools that can help you gain actionable insights. However, it is important to choose the right ones based on your company’s objectives.
Information theory recognises that all measurements are statistical in nature and thus always involve some uncertainty. It therefore defines measurement as a set of observations that reduce this uncertainty.
A unit is a standard quantity used to express the measurement of a physical quantity. For example, the length of a pencil is measured in units of “pencil length” so that it is easy to compare the lengths of different objects. Units can be combined to create derived units that measure more complex physical quantities. The International System of Units (SI) includes seven base units, including the metre for length, second for time, kilogram for mass, kelvin for temperature, candela for light, and mole for the number of particles in a sample.
The SI also uses a common set of prefixed units to denote multiples and fractions of the base units. This makes it easy to perform arithmetic calculations and solve application problems with any of the SI base units. The SI is not the only system of measurement, however, and several other systems are still in use. These other systems often have their own specific unit prefixes that are not included in the standard SI.
Uncertainty refers to the fluctuations in a measurement that result from random errors. These may include readings from different instruments, environmental changes and effects or operator error.
Inaccurate measurements are a reality in any business. The degree and types of inaccuracies that exist must be considered as part of a data analysis to ensure sound business decisions. Learning to calculate uncertainty can help businesses manage the uncertainty present in their measurements.
When calculating an uncertainty, it is important to consider both systematic and random error. Although most laboratory reports will only quote a combined standard uncertainty which represents the combination of both Type A and Type B uncertainties, it is important to realise that this value contains both random and systematic components. To obtain a more robust representation of uncertainty, these combined uncertainties should be multiplied by a coverage factor to produce an expanded uncertainty. This provides an estimate of the range in which the true quantity value may lie within a stated coverage probability or level of confidence.
While the words accuracy and precision are often used interchangeably, they have different meanings when referring to measurement. Accuracy describes how close a single measurement is to its true value, like a bullet hitting the center of a target, while precision explains how well a series of measurements agree with each other. This is why scientists typically report their values to a certain number of significant figures: this implies both an accuracy and a precision.
ISO defines accuracy as ‘the proximity of measurement results to their true value’ and precision as ‘the uniformity of repeated measurements under unchanged conditions’. Thus, a high level of accuracy requires both a low bias and a low variability (random error) while a high level of precision requires only a low bias.
A measure’s relevance to a particular problem depends on whether it conveys empirically significant information about that problem. This information could include the presence of an error in a measurement, the degree to which two quantities are alike, or the relationship between a quantity and another quantity.
The relevance of a measure is also influenced by the context in which it is used. In some cases, measurements are made to satisfy specific epistemic desiderata, such as the consistency and coherence of scientific theories. Other times, measurements are made to ensure that a product meets quality standards or to control the operations of an industry.
The National Institute of Standards and Technology, for example, sets standards for the metric system in the United States and regulates commercial measurements. These laws prevent fraud by requiring accurate records of the weight, volume and chemical composition of products. Measurements have pervasive effects on our daily lives. For example, a pilot checks his altimeter as he lands an airplane, and a driver glances at her speedometer.