Why nowcasting needs a tbl.now | The nowcasting workflow | The data | 1. Create the tbl_now | 2. Diagnose it | 3. Describe it | Summarise the data | Visualize the data | 4. Then: hand it over, or fit it here | Hand it over | Or fit it here | 5. Evaluate your nowcast | 6. And create ensembles | Where to go next | References

Why nowcasting needs a tbl.now | The nowcasting workflow | The data | 1. Create the tbl_now | 2. Diagnose it | 3. Describe it | Summarise the data | Visualize the data | 4. Then: hand it over, or fit it here | Hand it over | Or fit it here | 5. Evaluate your nowcast | 6. And create ensembles | Where to go next | References

Native modelling and the shared workflow | Your data: the tbl_now format | Example 1 -- Dengue fever (setting up a stratified nowcast) | Example 2 -- Mpox (modifying the nowcast model) | Example 3 -- Comparing models with a backtest | When a report is not yet a case: the revision process | Example 4 -- Letting the package choose the model (auto_nowcast()) | Updating the chosen model as new data arrive | Example 5 -- Count-cumulative revisions and a historical origin | Saving and loading a fitted nowcast | Combining diseasenowcasting models in an ensemble | Next steps

TL;DR | The challenge: few data points | 0) Prelude: Data preparation | 1) Prior choices at the start of an epidemic | 1.1 The SIR model: mechanistic priors | 1.2 The AR(1) model: trend priors | 1.3 The HSGP model: kernel priors | 2) Experiments with the prior: prior-predictive simulation | 2.1) SIR -- moving R0. | 2) COVID-19 Colombia: example from 1 day to 3 months | 2.2 Fitting each model across the four windows | 2.3 Comparing the three models across windows | 2.4 A rule of thumb | 3) Conclusion: When to fit what?

The one rule: RTMB | 1. Custom delay: Weibull | The data | The custom part: a Weibull delay | Fit it (and a built-in to compare against) | 2. Custom epidemic: different SIR | The custom part: an SIR epidemic | Fit it | 3. Summary: Tips for writing your own component | 4. Advanced: A differential equations epidemic process

Quick reference | What the rest of the vignette shows: | 1. Likelihood prior: NB precision (phi) | 2. Reporting-delay parameters | 2.1 Log-Normal delay | 2.1.1 Location (mu) | 2.1.2 Scale (sigma) | 2.2 Generalized-Gamma delay | 2.2.1 Location (mu) | 2.2.2 Scale (sigma) | 2.2.3 Tail shape (Q) | 2.3 Dirichlet (non-parametric) delay | 2.3.1 Concentration (alpha) | 3. Epidemic-process parameters | 3.1 HSGP | 3.1.1 Amplitude (alpha) | 3.1.2 Length-scale (ell) | 3.1.4 HSGP: number of basis functions (num_basis) | 3.2 AR(1) | 3.2.1 Autocorrelation (phi) | 3.2.2 Innovation SD (sigma) | 3.3 SIR | 3.3.1 Basic reproduction number (R0) | 3.3.2 Recovery rate (gamma) | 3.3.3 Susceptible fraction (N_eff) | 3.3.4 Total population (N_pop) | 4. Covariate priors (covariate_prior) | 5. Practical guide

Example 1: Estimating the PAF for the proportion of dementia attributable to smoking | Example 2: Combining subpopulations into a total | Example 3: Estimating the PIF for the proportion of dementia reduced by a 15% reduction in smoking. | Example 4: A categorical population attributable fraction | Example 5: Combining fractions of different risks into an ensemble | Example 6: Adjusted fractions | Example 7: Ensuring non-negative fractions | Example 8: Computing averted and attributable cases | Example 9: Computing averted cases (strictly positive) | Example 10: Using Hazard Ratios (HR) and Odds Ratios (OR) instead of Relative Risks (RR) | References

Core Concepts: PAF and PIF | Key assumption: Independent (summary) data sources. | Usage | Population Attributable Fraction (PAF) | Incorporating Uncertainty | Potential Impact Fraction (PIF) | Attributable and averted cases | Combining fractions from subpopulations | Combining fractions from multiple risks | Adjusting fractions for commonality | Additional information | References

Introduction | Comparing the Performance with and without DoW Effect | DoW Coefficients Estimates | Employing DoW Estimates as Priors for Future Nowcasts | Conclusion

Introduction | Nowcasting Scenarios | Comparing the Nowcast Estimates of the Poisson and negative binomial Models | Extracting Quantile Estimates | Calculating and Plotting the Weighted Interval Score | Conclusion

Introduction | Data Preparation | Nowcasting Scenarios | Delay Distribution Estimates | Incidence Estimates | Conclusion