Lesson 1Deployment considerations: retraining schedule, data pipelines, monitoring drift and backtesting forecastsPlan the rollout of revenue forecasting systems, including how often to retrain, automated data flows, watching for data and concept changes, and regular backtesting to keep performance steady in shifting market conditions like Nigeria's economy.
Building strong data pipelinesSetting retraining and update schedulesMonitoring data and concept driftContinuous backtesting and benchmarkingAlerting and rollback strategiesLesson 2Communicating forecasts to stakeholders: visualization of point forecasts and intervals, scenario analysis, transparency of assumptionsShare revenue forecasts clearly with stakeholders using simple charts for point and interval predictions, scenario reviews, and open notes on assumptions, limits, and model risks, especially for Nigerian business leaders.
Visualizing point and interval forecastsScenario and what-if analysis designExplaining drivers and key featuresDocumenting assumptions and limitsTailoring messages to stakeholdersLesson 3Machine learning time series models: random forests/gradient boosting with lagged features, XGBoost/LightGBM, and sequence models (LSTM/GRU)Train machine learning models for time series revenue prediction, including tree groups with past features and sequence models like LSTM and GRU, while managing nonstationarity, seasonal patterns, and differences across products in Nigerian banking.
Random forests with lagged featuresGradient boosting, XGBoost, LightGBMGlobal versus local forecasting modelsSequence models with LSTM and GRUHandling nonstationarity and scalingLesson 4Formulating forecasting objectives and evaluation horizons (e.g., next 3, 6, 12 months)Set forecasting goals for core bank products by choosing prediction targets, time frames, and detail levels, aligning them with business needs like budgeting, pricing, liquidity planning, and regulatory reporting in Nigeria.
Choosing revenue targets and unitsSelecting forecast horizons and frequencyAligning forecasts with business decisionsGranularity by product, segment, and regionHandling new products and short historiesLesson 5Identifying and sourcing time series data (public financial series, payment volumes, synthetic generation techniques)Learn to find, check, and gather time series data for bank revenue forecasting, including internal product stats, public financial data, and synthetic data that boosts limited or noisy historical records safely for Nigerian contexts.
Cataloging internal product revenue seriesUsing public macro and market data sourcesCollecting payment and transaction volume dataAssessing data quality, gaps, and revisionsSynthetic data generation for stress scenariosLesson 6Training and hyperparameter tuning: grid/random search, Bayesian optimization, time-aware scoringBoost model performance with planned hyperparameter tuning, like grid and random search, Bayesian methods, and time-sensitive scoring that follows time order and targets key business periods in Nigerian finance.
Defining search spaces and priorsGrid and random search trade-offsBayesian optimization workflowsTime-aware validation and scoringEarly stopping and resource limitsLesson 7Model ensembling and reconciliation: simple model averaging, weighted ensembles, stacking for time seriesMix multiple forecasting models for bank revenue using simple averages, weighted groups, and stacking, plus hierarchical reconciliation to keep forecasts consistent across products, branches, and levels in Nigerian banks.
Simple and weighted model averagingStacking and meta-learners for seriesDiversity and correlation among modelsHierarchical and grouped reconciliationEvaluating ensemble stability over timeLesson 8Baseline time series methods: ARIMA, ETS, naive and seasonal naïve models, decomposition (trend/seasonality)Check baseline time series models for bank revenue, like naive, seasonal naive, ARIMA, ETS, and decomposition, to set performance benchmarks and understand trends and seasons before advanced machine learning in Nigeria.
Naive and seasonal naive benchmarksClassical decomposition of trend and seasonalityARIMA modeling for bank revenue seriesExponential smoothing and ETS variantsComparing baselines across productsLesson 9Feature engineering for revenue: lags, rolling means/std, differencing, calendar effects, holiday indicators, cohort effects, marketing/campaign flagsCreate predictive features for bank revenue, including lags, rolling stats, differencing, calendar and holiday effects, cohort indicators, and marketing flags to catch demand shifts and breaks in Nigerian markets.
Lag and lead features for revenueRolling means, volatility, and ratiosCalendar, holiday, and payday effectsCohort and lifecycle based featuresMarketing and campaign impact flagsLesson 10Data splitting and cross-validation for time series: train/validation/test splits, expanding window CV, blocked CVSet up time-smart data splits and cross-validation for revenue forecasting, using rolling and blocked methods to prevent leaks, match real use, and get solid performance estimates over time in Nigerian data.
Holdout train, validation, and test splitsRolling and expanding window validationBlocked cross-validation for seasonalityPreventing temporal leakage in featuresBacktesting over multiple forecast originsLesson 11Evaluation metrics and error analysis: MAE, RMSE, MAPE, symmetric MAPE, prediction intervals and coverageAssess revenue forecasts with metrics like MAE, RMSE, MAPE, symmetric MAPE, and interval coverage, plus detailed error checks by segment, time frame, and conditions to spot biases and weaknesses in Nigerian finance.
Scale-dependent error metricsPercentage and relative error metricsPrediction intervals and coverageHorizon and segment level diagnosticsRegime and event-driven error analysisLesson 12Macro and external regressors: using CPI, unemployment, interest rates, mobility, Google Trends; feature selection and lag alignmentAdd macroeconomic and external factors to revenue models, like CPI, unemployment, interest rates, mobility, and search trends, with methods for lag matching, scaling, and selection to avoid overfitting in Nigeria.
Selecting relevant macro indicatorsAligning lags between macro and revenueTransforming and scaling external dataFeature selection and regularizationStress and scenario overlays with macros
