liveBookmanning.com

What you will learn in Effective Data Analysis

Hard and soft skills to accelerate your career

by Mona Khalil
Foreword by Barry McCardel

Read it on liveBook Get the book at manning.com

Chapter by chapter

Concepts explained in this book

• linear regressionLinearity is the property that a modeled relationship can be represented by additive changes proportional to predictor changes.
• metricA model evaluation metric is a quantitative measure of how well a statistical model fits data or performs its intended task.
• independent variableA predictor is an input variable whose relationship with an outcome is represented in a statistical model.
• descriptive statisticsDescriptive statistics are numerical or graphical methods for summarizing and characterizing the observed features of a dataset.
• granularityGranularity, or grain, is the level of detail represented by a record, observation, or dataset.
• correlationCorrelation is a statistical association describing how two variables vary together, without by itself establishing causation.
• self-report measureSelf-report data is information that people provide about their own behaviors, experiences, attitudes, or characteristics.
• language modelA large language model is a machine-learning model trained on extensive text data to process and generate human language.
• personal dataPersonal data, also called personal information or personally identifiable information, is information that identifies or can reasonably be linked to an individual.
• data sourceA data source is a system, repository, or collection from which data used for analysis or visualization is obtained.
• labelA class label is the categorical value assigned to an observation to identify its class.
• researchResearch is the systematic investigation of a topic through study or experimentation to obtain new information.
• anonymizationAnonymization, or data anonymization, transforms or removes identifying information so individuals cannot reasonably be identified from data.
• classifierA classification problem is a prediction task whose outcome is one of a set of discrete classes or categories.
• data explorationExploratory data analysis is the systematic examination of a dataset to understand its distributions, patterns, anomalies, and limitations before formal modeling or testing.
• normal distributionA normal distribution is a continuous probability distribution that is symmetric about its mean and has a bell-shaped density.
• parametric testA parametric statistical test is a statistical method whose validity depends on assumptions about a specified probability distribution characterized by a finite set of parameters.
• central tendencyA measure of central tendency is a statistic that represents a typical or central value in a dataset.
• goalA desired outcome that an organization seeks to achieve through its activities and decisions.
• data qualityData quality is the degree to which data is accurate, complete, consistent, timely, and fit for its intended analytical purpose.
• benchmarkA benchmark is a reference value or standard used to evaluate and compare an observed measurement or proportion.
• statistical significanceA property of a statistical result indicating that it would be relatively unlikely under a specified null hypothesis according to a chosen significance criterion.
• t-testA t-test is a statistical test that evaluates differences between means using a t-distributed test statistic.
• post hoc testA post hoc test is a follow-up statistical comparison performed after an omnibus test to identify which specific groups differ while accounting for multiple comparisons.
• operationalizationOperationalization is the process of translating an abstract or complex concept into a precisely defined, measurable procedure or indicator.
• coefficient of determinationR2 is the proportion of variation in an outcome explained by a fitted regression model relative to a baseline model.
• residual errorA residual is the difference between an observed outcome and the value predicted by a fitted model.
• behavioral dataUser behavior data is data recording the actions or interactions of users with a website, product, or service.
• contractAn API is a defined programmatic interface through which software requests data or operations from another software system or service.
• guardrail metricA counter metric is a measure monitored alongside a primary metric to detect harmful tradeoffs or unintended consequences.
• sensitive informationProtected information is data subject to legal, regulatory, or organizational restrictions governing its collection, use, disclosure, or retention.
• kruskal-wallis testThe Kruskal-Wallis test is a rank-based non-parametric statistical test for comparing three or more independent groups.
• p-valueA p-value is the probability of obtaining a test statistic at least as extreme as the observed value assuming the null hypothesis is true.
• ETLETL is the process of extracting data from sources, transforming it into a suitable structure, and loading it into a target storage system.
• percentage of leads qualified for salesPercentage of leads qualified for sales is the proportion of leads classified as meeting the criteria for sales qualification.
• analysis of varianceANOVA is a statistical method for testing whether means differ across multiple groups or conditions.
• SQLA SQL query is a statement written in SQL that retrieves, filters, combines, or modifies data in a database.
• friedman testThe Friedman test is a non-parametric rank-based test for comparing three or more related samples or repeated measurements.
• mann-whitney u testThe Mann-Whitney U test is a non-parametric statistical test for comparing the relative ranks or distributions of two independent samples.
• wilcoxon signed-rank testThe Wilcoxon signed-rank test is a non-parametric statistical test for comparing paired or repeated measurements using the ranks and signs of within-pair differences.
• data modelA data model is a defined representation of data entities, attributes, relationships, and organization for a particular analytical or operational purpose.
• experimentationAn experiment is a procedure that evaluates the effect of an intervention by comparing outcomes under different conditions.
• effect sizeAn effect size is a quantitative measure of the magnitude of a difference or relationship, independently of sample size.
• modeMode is an analytics platform that combines interactive notebooks, data querying, visualization, and report sharing.
• survey dataSurvey data is information collected from questionnaires or interviews about respondents’ opinions, perceptions, and self-reported behaviors.
• AICAIC is an information criterion that evaluates a statistical model by balancing goodness of fit against model complexity.
• BICThe Bayesian information criterion is a model-selection criterion that balances goodness of fit against model complexity using a sample-size-dependent penalty.
• documentationDocumentation is the systematic recording of methods, data, procedures, and decisions so that work can be understood, evaluated, or repeated.
• f-statisticAn F-statistic is a test statistic that compares explained variation with unexplained variation in an analysis of variance or regression model.
• homoscedasticityHomoscedasticity is the condition in which regression errors have a consistent spread across values of the independent variables.
• measurement scaleA rating scale is an ordered set of response options for indicating the degree or level of a characteristic, attitude, or judgment.
• t-valueA t-value is the test statistic from a t-test representing a standardized difference between means or between a sample mean and a hypothesized value.
• artificial intelligenceArtificial intelligence is the field of creating computational systems that perform tasks associated with human intelligence, such as reasoning, perception, language use, and generation.
• dimension tableA dimension table is a dimensional-modeling table containing descriptive attributes about entities or contexts associated with facts.
• out-of-sample forecastAn out-of-sample prediction is a model-generated estimate for an observation not included in the data used to fit the model.
• standard deviationStandard deviation is a measure of the typical spread of numerical observations around their mean.
• biasBias is a systematic pattern of error or unequal treatment that produces unfair or disadvantageous outcomes for particular groups.
• generative modelingA class of artificial-intelligence systems that generates new content such as text, code, images, audio, or video from learned patterns and inputs.
• meanThe mean is the arithmetic average of a set of numerical values.
• composite metricA composite measure or metric is a single score or indicator formed by combining multiple measurements or metrics.
• net promoter scoreThe Net Promoter Score is a standardized customer-loyalty metric calculated as the percentage of promoters minus the percentage of detractors.
• Cohen's dCohen's d is a standardized effect-size measure expressing the difference between two means in standard-deviation units.
• degree of freedomDegrees of freedom are the number of independent values available to estimate a statistical quantity or determine a test distribution.
• independenceIndependence is a condition in which the occurrence or value of one observation does not affect another observation.
• false positiveA Type 1 error is a false-positive conclusion that rejects a true null hypothesis or identifies an effect that is not present.
• self-serviceSelf-service analytics is a practice in which stakeholders independently explore, query, and interpret data using prepared analytical resources.
• informed consentInformed consent is a participant's voluntary agreement to take part in research after receiving sufficient information about its purpose, procedures, and potential risks.
• Likert scaleA Likert scale is an ordered response scale in which respondents indicate the degree to which they agree with or feel strongly about a statement or question.
• output formatA response format is the structure and set of options through which respondents provide answers to a question or assessment item.
• varianceVariance is the average squared deviation of observations from their mean and measures their dispersion.
• adjusted r-squaredAdjusted R-squared is a version of R-squared that penalizes the score for additional variables included in a model.
• validityValidity is the degree to which a measure accurately represents or supports conclusions about the construct it is intended to measure.
• predictive analyticsPredictive modeling is the construction and use of models to generate predictions for new or unknown observations.
• baselineA baseline is a reference measurement of a metric's current or historical performance before changes or interventions.
• false negativeA Type 2 error is a false-negative conclusion that fails to reject a false null hypothesis or fails to detect an effect that is present.
• medianThe median is the middle value of an ordered dataset, or the average of the two middle values when the dataset has even size.
• double-barreled questionA double-barreled question or statement combines two distinct issues in a single survey item, making a single response ambiguous.
• leading languageLeading language is wording that suggests a preferred answer or embeds assumptions about respondents.
• securityData protection is the safeguarding of personal information through controls on its collection, access, processing, storage, sharing, retention, and deletion.
• data frameA dataframe is a tabular data structure with labeled rows and columns that can hold variables of different types.
• eltELT is a data-integration process that extracts data, loads it into a target system, and transforms it after storage.
• interquartile rangeThe interquartile range is the spread of the middle 50 percent of observations between the 25th and 75th percentiles.
• standard errorStandard error is an estimate of the sampling variability of a statistic, such as the distance between a sample mean and the corresponding population mean.
• balanced Likert scaleA balanced Likert scale provides equal numbers of positive and negative response options.
• free-form textA free-text response is an unrestricted textual answer supplied in the respondent's own words.
• fully labeled scaleA fully labeled scale provides a verbal descriptor for the meaning of every response point.
• neutral midpointA neutral midpoint is a response option representing neither a positive nor a negative position on an ordered scale.
• partially labeled scaleA partially labeled scale provides verbal descriptors for only selected response points, typically endpoints and a midpoint.
• psychometricsPsychometrics is the field concerned with the theory and measurement of psychological and behavioral constructs.
• question wordingQuestion wording is the choice and arrangement of language used to formulate a survey question.
• response biasResponse bias is systematic distortion in responses caused by question wording, response options, context, or respondent characteristics.
• single-response questionA single-response question permits a respondent to select exactly one answer from a defined set of options.
• data aggregationAggregation is the process of combining multiple observations into summary values over groups or time periods.
• seasonality effectSeasonality is a recurring pattern in a variable associated with particular calendar periods or cycles.
• customer activity rateCustomer activity rate is the proportion of a defined period during which a customer performs a specified activity.
• prompt engineeringPrompt engineering is the practice of designing and refining instructions supplied to generative models to obtain desired outputs.
• SMART frameworkThe SMART framework is a goal- and metric-design approach requiring objectives to be specific, measurable, achievable, relevant, and time-bound.
• analytics engineeringAnalytics engineering is the function of developing and maintaining transformed data models and analytical data infrastructure.
• survival analysisSurvival analysis is a statistical approach for analyzing the timing of events while accounting for observations whose events have not yet occurred.
• scrapingWeb scraping is the automated retrieval and extraction of information from web pages, typically by requesting and interpreting their HTML content.
• intervalThe range is the difference between the highest and lowest values in a dataset, sometimes reported together with both endpoints.
• skewSkewness is the degree and direction to which a distribution is asymmetric around its center.
• time seriesA time series is a sequence of observations ordered or indexed by time points or intervals.
• study designThe planned structure of a study specifying how questions, data, methods, and comparisons will support valid conclusions.
• regular expressionA regular expression is a pattern notation for searching, matching, and extracting text according to specified character or structural criteria.
• log-likelihoodLog-likelihood is the logarithm of the likelihood of observed data under a statistical model and can support comparisons of model fit.
• p-hackingP-hacking is the manipulation of analyses, data collection, or reporting choices to obtain a statistically significant p-value.
• coefficientA coefficient in a statistical result is a numerical parameter estimate that quantifies the contribution or association of a modeled term.
• requestsRequests is a Python library for making HTTP requests to web services and retrieving their responses.
• multicollinearityMulticollinearity is strong correlation among independent variables in a regression model.
• between-subjects designA between-subjects design compares outcomes across separate groups of participants exposed to different conditions.
• hierarchical modelA hierarchical model represents data whose observations are nested within groups and estimates patterns at multiple levels.
• time-series modelA time series model uses observations ordered over time to estimate or predict future values while accounting for temporal patterns.
• feature engineeringFeature engineering is the creation or transformation of variables from existing data to better represent processes relevant to a model.
• automated decision toolAn automated decision tool is a software system that applies rules or learned algorithms to make or support decisions without direct human determination for each case.
• customer satisfactionCustomer satisfaction is a customer's evaluation of how well an organization's product or service meets expectations.
• reliabilityReliability is the consistency or stability of a measurement across repeated observations, items, or conditions.
• confounderA confounding variable is an uncontrolled factor associated with both an explanatory variable and an outcome that can distort their apparent relationship.
• training datasetA training set is the portion of data used to fit the parameters of a machine learning model.
• data handlingData processing is the set of operations that converts raw data into a usable format for analytical purposes.
• opportunity sizingOpportunity sizing is the process of estimating the value, costs, and expected effects of pursuing a product-development opportunity.
• cumulative metricA cumulative metric measures the total of current and previous values up to a specified time period.
• incremental metricA metric view that reports changes or additions over successive intervals rather than a cumulative total.
• fact tableA fact table is a tabular structure that records measurable events or processes, typically with numeric measures and references to dimensions.
• star schemaA star schema is a dimensional database design with a central fact table connected directly to surrounding dimension tables.
• pseudonymizationPseudonymization is the reversible replacement of identifying data with pseudonyms or other identifiers that limit but do not prevent reidentification.
• metadata catalogA data catalog is an inventory and discovery system that describes available data, its relationships, its provenance, and its ownership.
• privacyPrivacy is the ability of people to control access to information about themselves and how that information is collected or used.
• hypothesisA hypothesis is a testable proposition about a population, relationship, or data-generating process.
• primary keyA primary key is a field or combination of fields that uniquely identifies a record in a dataset or table.
• analystAn analyst is a professional who examines data to generate insights and support decisions.
• sample sizeSample size is the number of observations or participants included in a statistical sample.
• JSONJSON is a text-based data-interchange format that represents structured values using objects, arrays, and primitive values.
• measurementMeasurement is the systematic assignment of values or categories to attributes, phenomena, or processes so they can be represented and analyzed.
• knowledge managementKnowledge management is the creation, organization, and sharing of information within an organization.
• explanatory statistical modelAn explanatory model represents relationships between variables and an outcome with emphasis on interpretability and evaluation of theoretical effects.
• heteroscedasticityHeteroscedasticity is the condition in which the spread of residuals varies across the range of predicted or explanatory values.
• ordinary least squaresAn ordinary least squares regression model estimates coefficients by minimizing the sum of squared differences between observed and predicted outcomes.
• data applicationA data app is an interactive application whose interface or outputs are driven by data and computational code.
• promptA prompt is user-provided input that instructs a generative model about the content or behavior expected in its output.
• bootstrap samplingBootstrapping is a resampling method that repeatedly samples with replacement from observed data to estimate the variability of a statistic or inferential result.
• covariateA covariate is a variable included in or considered by a statistical model because it may vary with a predictor or outcome.
• statistical errorA statistical error is a discrepancy between a population value and an inference or estimate derived from a sample.
• root mean squared errorRoot mean squared error is the square root of the mean squared differences between predicted and actual numerical values.
• data warehouseA data warehouse is a centralized storage system that integrates and organizes large volumes of data for retrieval and analysis.
• power analysisAn a priori power analysis determines the sample size required to achieve a desired probability of detecting an effect of a specified magnitude at a chosen significance level.
• confidence intervalA confidence interval is an interval estimate constructed by a statistical procedure to represent plausible values for a population parameter at a specified confidence level.
• repeated measures t-testA repeated measures t-test is a statistical test that compares means from paired or otherwise dependent observations measured repeatedly.
• dimension reduction modelA dimension reduction model replaces many original variables with fewer composite dimensions that preserve much of their variability.
• hypothesis spaceA model class is a group of formulas and algorithms sharing characteristics and designed to solve a specific type of problem.
• data transformationA data transformation is a mathematical or computational modification of data values intended to change their representation, scale, distribution, or modeled relationship.
• causationCausation is a relationship in which one event, condition, or behavior produces a change in another.
• vanity metricA vanity metric is a measure that appears favorable or celebratory but generally provides little actionable insight.
• clickstreamClickstream data is a record of users’ clicks, navigation, and related activity on a website or application.
• customer support recordCustomer support data records interactions between customers and an organization's support function.
• event-level dataEvent-level data contains one record for each individual occurrence, such as a page view, click, or call.
• line plotA chart is a visual representation of data designed to communicate patterns, comparisons, or trends.
• paginationPagination is the division of a large result set into sequential pages returned through separate requests.
• rate limitRate limiting is a restriction on the number or frequency of requests that a service accepts during a specified period.
• logistic regressionLogistic regression is a statistical model for estimating the relationship between a categorical outcome and one or more predictors.
• auditAn audit is a systematic examination of a system, process, or output against defined requirements or evaluation criteria.
• automated systemAn automated system is a computational system that performs tasks or makes decisions with limited or no direct human execution at the time of operation.
• deep learningDeep learning is a machine-learning approach that uses multilayered neural networks to learn representations and patterns from data.
• disproportionate accuracyDisproportionate accuracy is unequal predictive accuracy across subgroups affected by an automated system.
• disproportionate outcomeA disproportionate outcome is an unequal consequence or service result experienced by different subgroups of people affected by a decision system.
• support ticketA support ticket is a customer-submitted record describing a request, problem, inquiry, or feedback item requiring service handling.
• protected demographicA protected class is a legally recognized demographic group whose members receive protection against specified forms of discrimination.
• rule-based systemA rules-based model produces outputs by applying explicitly specified logical rules, conditions, or keyword matches to input data.
• inter-rater reliabilityInter-rater reliability is the degree of agreement or consistency among independent raters evaluating the same subjects or events.
• internal consistencyInternal consistency is the degree to which items in a measurement instrument produce consistent responses because they assess a common construct.
• test-retest reliabilityTest-retest reliability is the consistency of a measure when the same individuals are assessed repeatedly over time while the measured construct remains stable.
• stakeholder interviewA stakeholder interview is a structured conversation that elicits domain context, rationale, expectations, or evidence from people affected by analytical work.
• open datasetOpen data is publicly accessible data that can be reused, analyzed, and redistributed with minimal access restrictions.
• metadataMetadata is data that describes the properties, context, provenance, or structure of other data.
• categorical dataA categorical variable is a variable whose values represent membership in distinct categories rather than measured numerical amounts.
• ordinal dataOrdinal data consists of observations assigned to categories with a meaningful order but without necessarily equal spacing between categories.
• deanonymizationA deanonymization algorithm is a computational procedure that links supposedly anonymous records to identifiable individuals using auxiliary information.
• indirect measureIndirect measurement assesses a phenomenon through respondents' perceptions or proxy indicators rather than directly observing the phenomenon itself.
• anonymized production dataAnonymized data is data modified so that individuals cannot be identified directly or reasonably linked to the records.
• CCPAThe California Consumer Privacy Act (CCPA) is a California privacy law granting consumers rights concerning businesses' collection, use, sharing, and sale of personal information.
• data privacy regulationA data protection regulation is a law governing how organizations collect, store, process, use, retain, and delete information about individuals and how individuals can exercise rights over that information.
• GDPRThe General Data Protection Regulation (GDPR) is a European Union regulation governing the processing and protection of personal data and related individual rights.
• HIPAAThe Health Insurance Portability and Accountability Act is a United States law that regulates the privacy and security of protected health information.
• right to erasureThe right to be forgotten is a privacy right allowing an individual to request deletion of personal data held by an organization under applicable law.
• sector-specific lawA sector-specific law is a regulation that governs data practices within a particular industry or field because of the nature and risks of its information.
• demographic informationDemographic information is data describing characteristics of people or user populations, such as location or behavioral background.
• foundational modelA foundation model is a broadly pretrained model that can be adapted or prompted for many downstream tasks and domains.
• hallucinationA hallucination is generated model output that is nonsensical, fabricated, or factually inaccurate despite being presented as a response.
• batch inferenceBatch prediction generates predictions for grouped records at scheduled intervals rather than individually on demand.
• scikit-learnscikit-learn is a Python library providing algorithms and utilities for data preparation, predictive modeling, and machine learning.
• evaluation datasetA validation set is a dataset reserved for assessing or tuning a model after training and before final evaluation or deployment.
• pipelineA data pipeline is an automated sequence of processing steps that retrieves, transforms, and delivers data between systems.
• data breachA data breach is an incident in which protected or sensitive data is accessed, disclosed, altered, lost, or acquired without authorization.
• computational notebookA computational document or environment that combines executable code, explanatory text, and dynamically generated interactive output.
• chi-square testA chi-square test is a statistical test that evaluates categorical data by comparing observed frequencies with frequencies expected under a specified hypothesis.
• data collectionData collection is the systematic process of gathering, capturing, and recording observations, measurements, or other information for storage or analysis.
• Beautiful SoupBeautifulSoup is a Python library for parsing HTML and XML documents into navigable object trees.
• quasi-experimental designA quasi-experiment is a study that evaluates differences or effects when the key independent variable is an inherent participant characteristic that cannot be randomly assigned or directly manipulated.
• measurement biasMeasurement bias is systematic error introduced when the measurement process produces responses that differ from the true values or construct of interest.
• self-selection biasSelf-selection bias is systematic distortion caused by differences between people who choose to participate in a study and those who do not.
• data dictionaryA data dictionary is documentation that defines a dataset’s fields, data types, valid response options, and related descriptive information.
• government data sourceA government data source is a publicly published collection or repository of data produced or maintained by a governmental agency.
• peer-reviewed research paperPeer-reviewed research is scholarly research whose methods, findings, and conclusions have been evaluated by subject-matter experts before publication.
• confidentialityConfidentiality is the protection of information from unauthorized access, use, or disclosure.
• informed consent documentAn informed consent document is a formal written document that gives prospective research participants the information needed to decide whether to participate.
• National Weather Service APIThe National Weather Service API is a government web API that provides weather forecasts and related meteorological data.
• scatter plotA scatterplot is a graph that displays paired numerical observations as points to reveal relationships between variables.
• binary featureA binary variable is a variable with exactly two possible categories or values.
• dummy codingDummy coding represents a categorical variable as separate binary indicator variables for its categories.
• schemaA schema is a logical organization of related database objects such as tables.
• probability distributionA distribution describes how values of a variable are arranged across their possible range and frequencies.
• retention policyA data retention policy specifies how long data may be kept and the conditions under which it must be anonymized or deleted.
• statistical modelingA statistical model is a mathematical representation of relationships among variables and random variation in observed data.
• database tableA table is a structured data collection arranged in rows representing records and columns representing attributes.
• visualizationData visualization is the graphical representation of data to communicate patterns, comparisons, or relationships.
• hash valueA hash code is the fixed-size output produced by applying a hash function to data.
• hash functionA hash algorithm is a computational procedure that deterministically maps data to a fixed-size hash code.
• hashlibhashlib is Python's standard library module providing common secure hash and message-digest algorithms.
• SHA-256SHA-256 is a member of the Secure Hash Algorithm 2 family that produces a 256-bit cryptographic hash.
• data lakeA centralized repository that stores large volumes of raw data in structured, semi-structured, or unstructured form for later processing and analysis.
• organizational strategy metricAn organizational strategy metric is a quantitative measure of a factor relevant to organizational performance and strategic decision-making.
• proactive analyticsProactive analytics is the practice of anticipating recurring information needs and preparing reusable analyses before individual requests arise.
• anonymized IDAn anonymized user ID is an identifier assigned to represent a user without directly storing the user's identifying attributes.
• non-parametric testA non-parametric statistical test evaluates evidence without requiring a specified parametric distributional form for the data.
• construct validityConstruct validity is the degree to which a measure accurately assesses the theoretical construct it is intended to measure and exhibits theoretically expected relationships.
• content validityContent validity is the degree to which a measure comprehensively captures all relevant facets of the concept or process it is intended to measure.
• criterion validityCriterion validity is the degree to which a measure is associated with an external criterion representing an established measure or outcome of interest.
• longitudinal study designA longitudinal study design assesses participants repeatedly at multiple points over time to identify changes or trends.
• within-subjects designA within-subjects design compares conditions using the same participants, often by measuring them before and after treatment.
• descriptionA descriptive method summarizes observed characteristics of data without testing an intervention or manipulating a condition.
• structured dataStructured data is information organized according to a predefined format that enables consistent storage, retrieval, and processing.
• Cronbach's alphaCronbach's alpha is a coefficient estimating the internal consistency of a collection of measurement items from their inter-item covariance or correlation.
• pingouinpingouin is a Python statistical package providing common statistical tests and comprehensive results, including internal-consistency analysis.
• dashboardA dynamic dashboard updates or responds interactively to enable exploration of data and analysis results.
• random group assignmentRandom assignment allocates study participants to experimental conditions by a chance-based procedure.
• qualitative analysisQualitative analysis is the synthesis, interpretation, and examination of non-numerical information.
• trendA trend is a sustained pattern of change in a variable across an ordered period.
• foreign keyA foreign key is a column or set of columns whose values reference a key in another table to represent a relationship between rows.
• data minimization strategyData minimization is the practice of limiting the collection, storage, and use of personal data to what is relevant and necessary for a specific purpose.
• accountability metricAn accountability metric is a quantitative measure used to evaluate the responsibilities, performance, or outcomes attributable to a person or team.
• data infrastructureData infrastructure is the collection of systems, tools, and processes that gathers, stores, processes, retrieves, and presents organizational data for analysis.
• explicit user consentConsent is an informed and voluntary permission for an organization to collect, access, share, or use a person's data for a specified purpose.
• database normalizationDatabase normalization is the process of organizing relational data to reduce duplication and dependency anomalies.
• denormalizationDenormalization is the deliberate combination or duplication of related data to reduce separation between tables and simplify analytical access.
• residual plotA residual plot is a graph of model residuals against predicted values or explanatory-variable values used to assess error patterns.
• analytics telephoneA communication failure in which analytical results are passed between people or teams and acquire conclusions beyond the scope of the original work.
• document databaseA document database is a database that stores complex nested records, typically in document formats with flexible or dynamic structures.
• non-significant differenceA non-significant difference is an observed difference for which the available statistical evidence does not cross the selected significance threshold.
• cohort analysisA cohort study design follows a defined group of individuals sharing specified characteristics to observe outcomes or changes.
• prototypingRapid prototyping is an approach that creates and iteratively refines preliminary versions of a product or deliverable.
• StreamlitStreamlit is an open-source Python framework for building and sharing interactive data applications.
• controlled environmentA controlled environment is a setting in which potential explanatory conditions are managed or removed to isolate the effect of a variable.
• evidenceEvidence is information that supports or challenges a hypothesis, theory, or conclusion without making it irrevocable.
• mathematical proofProof is a claim treated as an established rule rather than as a conclusion open to refutation by future evidence.
• samplingRandom selection is a sampling procedure in which participants or subjects are chosen from a population using chance-based selection.
• data-informed cultureA data-informed culture is an organizational environment in which decisions and interpretations are shaped by appropriately evaluated data and analytical evidence.
• guardrailA guardrail is a predefined criterion or control intended to maintain analytical accuracy, protect individuals, and constrain harmful or unreliable interpretations.
• scope of interpretationThe scope of interpretation is the degree to which analytical findings can reasonably be generalized beyond the people included in the analysis.
• multiple-comparisons problemMultiple comparisons are simultaneous or sequential statistical tests involving several groups, hypotheses, or pairwise contrasts.
• bar chartA bar graph is a chart that represents values with bars whose lengths or heights encode quantitative comparisons across categories.
• dual-axis chartA chart configuration that displays two vertical axes with separate scales for different metrics.
• cumulative viewA representation that displays a running total accumulated across successive time periods or observations.
• incremental viewA representation that displays the value or change attributable to each individual time period or observation.
• logarithmic scaleA numerical axis scale in which equal visual intervals represent equal multiplicative ratios.
• user engagementWebsite engagement is the extent and quality of users’ interactions with a website, such as viewing pages, spending time in sessions, or clicking interface elements.
• accuracyData accuracy is the degree to which data correctly represents the real-world process or phenomenon it claims to describe.
• completenessCompleteness is the degree to which all required data is present for the intended analysis or operation.
• consistencyConsistency is the degree to which data conforms to defined rules, constraints, formats, and representations across records and sources.
• timelinessTimeliness is the degree to which data is current and available when it is needed.
• expected countAn expected frequency is the counted or proportional occurrence predicted for a category under a specified null hypothesis.
• goodness-of-fit testA goodness-of-fit test is a chi-square test that compares observed category frequencies in one sample with expected frequencies.
• observed countAn observed frequency is the counted or proportional occurrence of a category in the collected data.
• statistical independence testAn independence test is a chi-square test that evaluates whether the distributions of two categorical variables are associated.
• trend testA trend test is a chi-square test that evaluates whether category frequencies change systematically across ordered groups or time.
• convergent validityConvergent validity is the degree to which a measure is associated with other measures that theoretically assess the same construct.
• discriminant validityDiscriminant validity is the degree to which a measure is not associated with measures of theoretically unrelated constructs.
• predictive validityPredictive validity is the degree to which a measure is associated with a criterion assessed at a subsequent time.
• data organizationData structuring is the process of cleaning, transforming, and organizing raw information into structures that an organization can readily use.
• dbtdbt is an open-source tool that enables analytics engineers to transform data in a warehouse using SQL-based projects, testing, documentation, and software-development practices.
• join keyA join key is a field whose matching values identify records that can be combined across data structures.
• data typeA data type specifies the kind of value a field can contain and the operations or representation associated with that value.
• anonymityAnonymity is the condition in which an individual's identity cannot be linked to their data or actions.
• contextContextual information is a combination of non-sensitive details that provides enough clues to infer an individual's identity.
• minimum aggregation levelA minimum aggregation level is the smallest permitted number of individuals represented in a reported aggregate subgroup.
• data analysisAnalytics is the practice of using data to discover and communicate patterns, trends, and insights that inform decisions.
• key performance indicatorA key performance indicator is a metric selected to evaluate progress toward an important organizational objective.
• HTMLHTML is a markup language used to structure content in documents displayed by web browsers.
• robots.txtrobots.txt is a convention in which a website publishes crawler-access rules in a file at a standard location.
• employee engagementEmployee engagement is the degree of an employee’s involvement, commitment, and connection to their work and organization.
• parallelismParallel processing is the simultaneous execution of separate portions of a computational task.
• computer clusterA computing cluster is a group of networked computers that jointly execute computational tasks.
• distributed programmingDistributed computing is the execution of computational work by dividing it across multiple networked computers that process tasks in parallel.
• generalizationThe extent to which findings from a study apply to populations, settings, or conditions beyond those directly examined.
• limitations sectionA report or presentation section that states unavailable data, interpretive boundaries, and conclusions that the findings cannot support.
• precision of languageThe deliberate use of precise analytical terminology to communicate the warranted interpretation of findings.
• filteringA data-selection operation that restricts displayed or analyzed records according to specified criteria.
• codebookA codebook is documentation that describes a dataset's variables, coding schemes, methodology, and usage information.
• data portalA data portal is an interactive web interface for discovering, filtering, aggregating, visualizing, and downloading data.
• endpointAn API endpoint is a specific network-accessible address at which an API receives requests for a particular resource or operation.
• information schemaAn information schema is a set of database metadata views describing objects such as schemas, tables, columns, and their properties.
• program evaluationProgram evaluation is the systematic assessment of a program’s effectiveness in achieving its intended goals for a target population.
• ethicsEthics in science concerns principles governing responsible scientific practice and the treatment of people, animals, and society.
• bayesian statisticsBayesian statistics is an approach to statistical inference that updates probability distributions for hypotheses or parameters using observed evidence.
• frequentist hypothesis testFrequentist hypothesis testing is an inferential framework that evaluates hypotheses by considering the behavior of sample-based results across repeated samples from a population.
• hypothesis quantificationHypothesis quantification is the process of specifying the expected numerical magnitude of a hypothesized difference or change.
• probability theoryProbability is a mathematical measure of uncertainty about events or propositions.
• profit marginProfit margin is the proportion of a product's revenue remaining as profit after relevant costs are accounted for.
• reportA report is a structured communication artifact that presents analyzed information and findings.
• human subjectA human subject is a living person about whom a research investigator obtains data through intervention, interaction, or identifiable private information.
• viewA view is a virtual tabular result defined by a query over one or more underlying tables.

These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.