We define a linguistic distribution as the range of values for a quantitative linguistic variable across the texts in a corpus. An accurate parameter estimate means that the measures based on the corpus are close to the actual values of a parameter in the domain. Precision refers to whether or not the corpus is large enough to reliably capture the distribution of a particular linguistic feature. Distribution considerations relate to the question of how many texts are needed. The answer will vary depending on the nature of the linguistic variable of interest. Linguistic variables can be categorized broadly as linguistic tokens (rates of occurrence for a feature) and linguistic types (the number of different items that occur). The distribution considerations for linguistic tokens and linguistic types are fundamentally different. Corpora can be “undersampled” or “oversampled” – neither of which is desirable. Statistical measures can be used to evaluate corpus size relative to research goals – one set of measures enables researchers to determine the required sample size for a new corpus, while another provides a means to determine precision for an existing corpus. The adage “bigger is better” aptly captures our best recommendation for studies of words and other linguistic types.

This chapter presents a general overview of sensor characterization from a system perspective, without any reference to a specific implementation. The systems are defined on the basis of input and output signal description and the overall architecture is discussed, showing how the information is transduced, limited, and corrupted by errors. One of the main points of this chapter is the characterization of the error model, and how this one could be used to evaluate the uncertainty of the measure, along with its relationship with resolution, precision and accuracy of the overall system. Finally, the quantization process, which is at the base of any digital sensor systems, is illustrated, interpreted, and included in the error model.

The Saudi-led intervention in Yemen is a valuable case study in the coercive use of air power. Saudi Arabia’s bombing campaign demonstrates the danger of employing a punishment approach against a subnational actor in a multi-sided internal conflict. Strategies of collective punishment, blockade, and decapitation have all malfunctioned against a stubborn and resilient Houthi adversary. The early audit from Yemen endorses a denial strategy, supports the growing orthodoxy that air attack is most effectively applied in support of ground forces, and offers insight on the relative utility of interdiction and close air support for that purpose. The Saudi-led coalition’s performance also underscores how difficult it is to achieve positive objectives with proxy warfare, regardless of air support. This chapter dissects the campaign, assesses its effectiveness, and draws lessons about air power’s ability to influence the outcome of similar complex civil war scenarios elsewhere.

Moscow’s air power success in Syria presents an opportunity to assess Russian inter- and intra-war adaptation in kinetic counterinsurgency. New technologies and tactics have enhanced the Russian Aerospace Force’s battlefield lethality and resilience but have not yet triggered a fundamental transition in operating concept. Russia’s air force has yet to actualize a reconnaissance-strike regime or advanced air-ground integration. Instead, situational and strategic factors appear to be more powerful contributors to its superior performance in the Syrian conflict. The way in which Russia has chosen to leverage its improvements in accurate munitions delivery, moreover, highlights key differences between its warfighting philosophy and that embraced by major Western powers. The resultant findings provide insight into Moscow’s coercive campaign logic, force-planning imperatives, and the likelihood that it might re-export the Syria model elsewhere.

As an empirical science, the study of animal behaviour involves measurement. When an animal engages in a series of actions, such as exploring or catching prey, the problem becomes that of identifying suitable components from that stream of action to use as markers suitable to score. The markers selected reflect the observer’s hypothesis of the organisation of the behaviour. Unfortunately, in most cases, researchers only provide heuristic descriptions of what they measure. To make the study of animal behaviour more scientific, the hypotheses underlying the decision of what to measure should be made explicit so as to allow them to be tested. Using hypothesis testing as a guiding framework, several principles that have been shown to be useful in identifying behavioural organisation are presented, providing a starting point in deciding what markers to select for measurement.

Poor-quality measurements are likely to yield meaningless or unrepeatable findings. High-quality measurements are characterised by validity and reliability. Validity relates to whether the right quantity is measured and is assessed by comparing a metric with a gold-standard metric. Reliability relates to whether measurements are repeatable and is assessed by comparing repeated measurements. The accuracy and precision with which measurements are made affect both validity and reliability. A major source of unreliability in behavioural data comes from the involvement of human observers in the measurement process. Where trade-offs are necessary, it is better to measure the right quantity somewhat unreliably than to measure the wrong quantity very reliably. Floor and ceiling effects can make measurements useless for answering a question, even if they are valid and reliable. Outlying data points should only be removed if they can be proved to be biologically impossible or to result from errors.

This chapter starts with basic definitions such as types of machine learning (supervised vs. unsupervised learning, classifiers vs. regressors), types of features (binary, categorical, discrete, continuos), metrics (precision, recall, f-measure, accuracy, overfitting), and raw data and then defines the machine learning cycle and the feature engineering cycle. The feature engineering cycle hinges on two types of analysis: exploratory data analysis, at the beginning of the cycle and error analysis at the end of each feature engineering cycle. Domain modelling and feature construction concludes the chapter with particular emphasis on feature ideation techniques.

This chapter describes how the world’s first independent air force, led by Air Chief Marshal Sir Hugh Trenchard, reacted to the threats to its existence by maximizing the Royal Air Force’s (RAF) operational utility and financial efficiency, while simultaneously contriving a credible narrative about its future strategic potential. In pursuing these twin narratives, the RAF developed a unique culture of beliefs and taken-for-granted attitudes that thrived because of the conceptually incurious nature of the men it selected to become officers. Few of these technically able "practical men" were willing to challenge their superiors’ intuitive and speculative belief that the morale of civilian populations was especially vulnerable to bombing. Instead, like their leaders, they became consciously complicit in acceding to the societal prophecies, articulated in books, films, and newspapers, that bombing would have apocalyptic effects, and that civil societies subjected to its effects would wish to sue for peace. The chapter concludes by analyzing how this culture impeded the realization that the anticipated outcomes were not being achieved and explains how this stymied options to pursue alternative strategies.

Two kinds of war have characterized the development of US Air Force culture. The first has involved struggles for air supremacy and decisive impact against a series of opponents, from Germany to Japan to Vietnam to Iraq. The second has involved bureaucratic fights against the US Army and the US Navy in the halls of Congress and the Pentagon. Combined, these fights have led to emphasis on recurring elements across the history of Air Force culture, including information precision, technological dominance, and decisive effect. These concentrations have structured how the US Air Force has fought its opponents, foreign and domestic, from before organizational independence in 1947 to the present day.

Errors in data are a part of life for experimenters in science and engineering. This chapter considers the types of errors, including random and systematic error that can occur during an experiment and methods by which uncertainties arising from such errors can be combined. Many worked examples are included in this chapter, as well as exercises for the student to complete