Data Analysis and Decision Making Training Course

Participants can identify suitable data sources and assess whether fields support the intended analysis.

• Data inventory records source systems, owners, refresh frequency and access controls
• Grain identifies what one row represents before joining datasets
• Data dictionaries document fields, definitions, units, formats and permitted values
• Primary keys, foreign keys and cardinality prevent duplicate counting
• ETL stages extract, transform and load data with auditability
• Sampling frames define population coverage and exclusion risks explicitly

Participants can assess data quality, document cleaning decisions and validate analysis-ready datasets.

• Completeness, validity, consistency, uniqueness, timeliness and accuracy quality dimensions
• Profiling distributions exposes missing values, impossible ranges and unexpected categories
• Missingness mechanisms distinguish MCAR, MAR and MNAR assumptions
• Outlier review combines IQR rules, z-scores and business context
• Reconciliation ties totals to source reports before downstream analysis
• Cleaning logs preserve original values, transformations, rationale and reviewer approval

Participants can prepare analysis-ready datasets using SQL and spreadsheet transformation techniques.

• SQL SELECT, WHERE, GROUP BY and HAVING answer scoped questions
• INNER, LEFT and FULL joins combine tables without inflating records
• Pivot tables summarise dimensions, measures, filters and calculated fields
• Date parsing standardises periods, fiscal calendars and rolling windows
• Derived variables calculate rates, ratios, cohorts and categorical bands
• Reproducible transformation steps separate raw, staged and analysis-ready datasets

Participants can describe patterns, variation and relationships using appropriate exploratory techniques.

• Frequency tables and cross-tabulations reveal distributions and segment differences
• Mean, median and mode suit different distribution shapes and decisions
• Variance, standard deviation and coefficient of variation quantify dispersion
• Percentiles and box plots identify spread, skewness and unusual values
• Index numbers track change while controlling base periods and units
• Correlation matrices identify candidate relationships without claiming causation

Participants can test business hypotheses and communicate statistical evidence with appropriate caution.

• Confidence intervals communicate estimate precision and sampling uncertainty clearly
• Null and alternative hypotheses structure statistically testable business questions properly
• t-tests compare group means under stated assumptions and sample conditions
• Chi-square tests assess association between categorical variables with contingency tables
• ANOVA compares multiple group means and controls familywise error
• Effect sizes distinguish practical importance from statistical significance

Participants can select, interpret and assess basic regression models for decision support.

• Simple linear regression estimates directional relationships between continuous variables
• Multiple regression controls observed factors and interprets coefficient estimates
• Logistic regression estimates event probabilities for binary outcomes
• Train-test splits assess generalisation beyond the development sample
• Confusion matrices report precision, recall, specificity and classification accuracy
• Overfitting checks compare training performance against holdout results

Participants can analyse customer groups, recurring patterns and operational trends over time.

• Time-series plots separate trend, seasonality, cycles and irregular variation
• Moving averages smooth short-term variation for operational forecasting
• Seasonal indices adjust recurring patterns before comparing periods
• Customer or beneficiary segmentation combines needs, value and behaviour
• RFM analysis prioritises customers by recency, frequency and monetary value
• Cohort analysis tracks retention, repeat activity and outcomes over time

Participants can present analysis through clear visualisations and evidence-based recommendations.

• Chart selection matches comparison, distribution, relationship, composition and time questions
• Bar, line, scatter and box plots avoid misleading encodings
• Axis scales, baselines and annotations prevent distorted visual interpretation
• Dashboard wireframes organise KPIs, filters, exceptions and drill-down paths
• Data storytelling links evidence, implications, options and recommended action
• Executive briefings state uncertainty, limitations and decisions requiring sponsorship

Participants can establish analytical controls and plan a workplace analysis from question to action.

• CRISP-DM structures business understanding, data preparation, modelling and deployment
• Data minimisation limits collection to stated analytical purposes
• Role-based access controls protect sensitive records and analytical outputs
• Model documentation records assumptions, inputs, performance and intended use
• Decision logs connect evidence, recommendations, approvals and measured outcomes
• Application plans define a priority question, dataset, method and stakeholder