R Programming for Statistical Computing Training Course

Participants can select and manipulate appropriate R data structures for analytical tasks.

• Vectors, matrices, lists, factors, and data frames compared
• Indexing observations and variables with numeric, logical, and named selectors
• Conditional expressions using if, else, case_when, and logical operators
• Iteration with for loops, while loops, apply functions, and map
• Factor levels, ordered categories, and reference groups for modelling
• Date-time classes, time zones, and duration calculations using lubridate

Participants can import common source files and assess data quality before analysis.

• CSV and delimited files imported with explicit encoding and column types
• Excel workbooks read from selected sheets, ranges, and named cells
• Inspecting dimensions, variable names, duplicate records, and unexpected values
• Data dictionaries documenting source fields, definitions, units, and permitted values
• Missing-data patterns assessed before exclusion, imputation, or recoding decisions
• Validation rules applied to identifiers, dates, ranges, and category values

Participants can create analysis-ready datasets using dplyr transformation pipelines.

• filter, select, arrange, mutate, and summarise applied within pipelines
• Grouped summaries calculated by customer, region, period, or programme
• Mutate statements creating rates, ratios, flags, and derived categories
• case_when rules translating business classifications into coded variables
• Inner, left, right, and full joins matched to reporting requirements
• Join diagnostics identifying unmatched keys, duplicated matches, and row inflation

Participants can reshape inconsistent data and prepare text and date fields for analysis.

• pivot_longer and pivot_wider reshaping repeated measures and reporting tables
• separate, unite, and extract methods splitting compound identifier fields
• String cleaning with regular expressions, pattern matching, and standardisation
• Date parsing from mixed formats, fiscal periods, and reporting timestamps
• Outlier flags built from business rules, quantiles, and z-scores
• Analysis variables engineered from raw transactions, events, and classifications

Participants can design clear ggplot2 charts that communicate analytical findings accurately.

• Grammar of graphics layers, aesthetics, geoms, scales, and themes
• Bar, line, scatter, histogram, boxplot, and density charts selected appropriately
• Faceting comparisons by region, customer segment, product, or time
• Colour palettes, labels, annotations, and legends designed for accessibility
• Logarithmic scales and transformations used without misleading visual interpretation
• Chart exports produced with controlled dimensions, resolution, and file formats

Participants can profile distributions, relationships and data-quality issues before formal modelling.

• Frequency tables, cross-tabulations, proportions, and cumulative distributions calculated accurately
• Mean, median, variance, standard deviation, and interquartile range compared
• Distribution shape assessed through histograms, density plots, and quantile plots
• Correlation matrices interpreted alongside scatterplots and substantive variable definitions
• Segment analysis comparing outcomes across groups with unequal sample sizes
• Exploratory findings recorded as testable questions and analytical assumptions

Participants can produce a documented analytical report from organised R project files.

• Project folders separating raw data, processed data, scripts, outputs, and documentation
• R Markdown documents combining narrative, code, tables, charts, and results
• Code chunks configured for readable output, warnings, messages, and caching
• Parameters used to rerun reports for different periods or business units
• Relative file paths protecting portability between analyst workstations and shared repositories
• Analytical assumptions, exclusions, and source references recorded within reports

Participants can select, run and interpret common inferential tests in R.

• Sampling distributions, standard errors, confidence intervals, and statistical uncertainty
• Null and alternative hypotheses stated for measurable business questions
• t-tests comparing means for independent, paired, and one-sample designs
• Chi-square tests assessing association between categorical variables and outcomes
• ANOVA comparing group means with planned post-hoc comparisons
• P-values, effect sizes, power, and practical significance interpreted together

Participants can build, interpret and diagnose multiple linear regression models.

• Linear regression equations linking continuous outcomes to explanatory variables
• Model formulas specifying main effects, interactions, transformations, and categorical predictors
• Coefficient estimates interpreted with confidence intervals and business units
• Residual plots checking linearity, constant variance, normality, and influential observations
• Multicollinearity assessed using correlation patterns and variance inflation factors
• Adjusted R-squared, RMSE, and holdout error used for model comparison

Participants can estimate binary-outcome models and evaluate classification performance.

• Binary outcomes modelled with binomial generalised linear regression
• Odds ratios and predicted probabilities translated into decision-relevant language
• Classification thresholds selected using costs, prevalence, and operational capacity
• Confusion matrices measuring sensitivity, specificity, precision, and overall accuracy
• ROC curves and area under curve comparing classification models
• Calibration checks comparing predicted risk with observed event rates

Participants can quantify model uncertainty and assess performance using resampling methods.

• Train-test splits preserving outcome proportions and temporal ordering where needed
• K-fold cross-validation estimating out-of-sample model performance
• Bootstrap samples calculating confidence intervals for uncertain statistics
• Monte Carlo simulation testing assumptions, scenarios, and decision thresholds
• Random seeds controlling repeatable sampling and simulation results
• Data leakage identified between training, validation, scoring, and reporting datasets

Participants can prepare time-indexed data and produce baseline forecasts with diagnostics.

• Regular time indexes constructed from daily, monthly, quarterly, and annual records
• Trend, seasonality, cycles, and irregular variation identified graphically
• Lagged variables, moving averages, and growth rates calculated correctly
• Autocorrelation and partial autocorrelation examined before model selection
• ARIMA models fitted using differencing, autoregressive, and moving-average terms
• Forecast accuracy measured with MAE, RMSE, MAPE, and forecast bias

Participants can retrieve, combine and document data from databases and web APIs.

• DBI connections querying relational databases without exporting unnecessary records
• SQL queries parameterised for filters, joins, aggregations, and reproducible extracts
• REST API requests constructed with endpoints, parameters, headers, and authentication
• JSON responses parsed into tabular structures for subsequent analysis
• Pagination, rate limits, retries, and failed requests handled predictably
• Source provenance recorded for downloaded data, query dates, and refresh rules

Participants can create maintainable R code with reusable functions and review controls.

• Custom functions defined with arguments, defaults, validation, and returned objects
• Functional programming applied to repeated files, groups, and reporting periods
• Unit tests checking expected outputs, boundary cases, and error conditions
• Assertions detecting impossible values before downstream calculations proceed
• Git commits recording meaningful changes to scripts, reports, and documentation
• Code review criteria covering readability, correctness, reproducibility, and assumptions

Participants can build, review and present a complete reproducible R analysis addressing a defined decision question.

• Decision question framed with stakeholders, measures, scope, and analytical constraints
• Raw data imported, validated, transformed, and documented in a project structure
• Exploratory analysis identifying patterns, anomalies, limitations, and candidate explanations
• Statistical model selected, validated, and interpreted against stated assumptions
• Visualisations and tables prepared for technical reviewers and decision makers
• Reproducible report presenting methods, findings, recommendations, and limitations