# Effect Prophet

> Forecasting for TypeScript

Give it your history. Get back a forecast you can explain, with the trend, weekly and yearly patterns, and holidays each broken out.

## Table of Contents

### Guide

- [What is Effect Prophet?](https://jer-k.github.io/effect-prophet/guide/introduction.md): What Effect Prophet forecasts, how it relates to Meta's Prophet, and when to use it.
- [Getting started](https://jer-k.github.io/effect-prophet/guide/getting-started.md): Install effect-prophet, fit a model to your history and forecast future dates.
- [Browsers and bundlers](https://jer-k.github.io/effect-prophet/guide/browsers.md): Use Effect Prophet in browsers with Vite, webpack, Rspack or esbuild, and how the WebAssembly engine loads.
- [Effect in five minutes](https://jer-k.github.io/effect-prophet/guide/effect-basics.md): The Effect concepts you need to call Effect Prophet: running effects, providing the fitting backend and handling typed errors.
- [Preparing your data](https://jer-k.github.io/effect-prophet/guide/your-data.md): The history and future row formats: timestamps, values, missing values, ordering, duplicates and how much history to use.
- [Country holiday calendars](https://jer-k.github.io/effect-prophet/guide/holiday-calendars.md): Add a built-in country holiday calendar with countryHolidays, and what its coverage does and doesn't guarantee.
- [Reading a forecast](https://jer-k.github.io/effect-prophet/guide/reading-forecasts.md): The fields predict returns for each forecast, including the trend and the additive and multiplicative components.
- [Tracing](https://jer-k.github.io/effect-prophet/guide/tracing.md): The spans and attributes Effect Prophet records for fits, forecasts and evaluations, and how to send them to a tracing backend.

### Examples

- [Trends and changepoints](https://jer-k.github.io/effect-prophet/examples/trend.md): Linear trends and changepoints: automatic and explicit changepoints, and how to make the trend more or less flexible.
- [Seasonality](https://jer-k.github.io/effect-prophet/examples/seasonality.md): Built-in yearly, weekly and daily seasonalities, custom periods, multiplicative effects and conditional seasonalities.
- [Holidays and events](https://jer-k.github.io/effect-prophet/examples/events.md): Model holidays and one-off events with custom event calendars and country holidays.
- [Extra factors (regressors)](https://jer-k.github.io/effect-prophet/examples/regressors.md): Add regressors such as temperature or ad spend, and forecast a regressor with its own model.
- [Growth with a ceiling](https://jer-k.github.io/effect-prophet/examples/saturating-growth.md): Logistic growth with a capacity, and an optional floor, for values that level off.
- [Flat trends](https://jer-k.github.io/effect-prophet/examples/flat-trend.md): Flat growth for data that moves around a steady level, with seasonality, events and regressors.
- [Uncertainty ranges](https://jer-k.github.io/effect-prophet/examples/uncertainty.md): Uncertainty ranges with predictWithIntervals, how they're simulated, and working with the raw samples.
- [Full Bayesian fitting](https://jer-k.github.io/effect-prophet/examples/bayesian.md): Full Bayesian fitting with the mcmc option, like Python's mcmc_samples: posterior sampling, ranges for every component, and sampler diagnostics.
- [Finding anomalies](https://jer-k.github.io/effect-prophet/examples/anomalies.md): Find unusual values with detectAnomalies for new measurements and anomalies for a whole history.
- [Checking the fit](https://jer-k.github.io/effect-prophet/examples/residuals.md): Check a fitted model with residuals, and check whether the L-BFGS optimizer stopped before the best fit.
- [Saving and loading models](https://jer-k.github.io/effect-prophet/examples/saving-models.md): Save and load fitted models as JSON, and refit from a warm start with warmStartParameters when new rows arrive.
- [Testing forecast accuracy](https://jer-k.github.io/effect-prophet/examples/cross-validation.md): Measure forecast accuracy with crossValidate and performanceMetrics, custom metrics and scaled errors.
- [Choosing settings](https://jer-k.github.io/effect-prophet/examples/tuning.md): Choose settings with comparePerformance and searchModels, compare against baselines and check on a final holdout.
- [Handling errors](https://jer-k.github.io/effect-prophet/examples/errors.md): The typed errors each operation can fail with, their reason values, and handling them with or without Effect.

### Python Prophet

- [Coming from Python](https://jer-k.github.io/effect-prophet/python/coming-from-python.md): Map Python Prophet's API to Effect Prophet: a side-by-side example and a table of equivalent options and methods.
- [How close are the results?](https://jer-k.github.io/effect-prophet/python/accuracy.md): How closely Effect Prophet matches Python Prophet 1.4.0, how that's checked, and why results aren't always identical.
- [What's different](https://jer-k.github.io/effect-prophet/python/differences.md): Where Effect Prophet differs from Python Prophet 1.4.0: intentional differences, smaller behavior differences and limits.

### Benchmarks

- [Benchmarks](https://jer-k.github.io/effect-prophet/benchmarks.md)
- [Linear growth benchmarks](https://jer-k.github.io/effect-prophet/benchmarks/linear.md)
- [Flat growth benchmarks](https://jer-k.github.io/effect-prophet/benchmarks/flat.md)
- [Logistic growth benchmarks](https://jer-k.github.io/effect-prophet/benchmarks/logistic.md)
- [Seasonality, events and regressors benchmarks](https://jer-k.github.io/effect-prophet/benchmarks/features.md)
- [Uncertainty ranges benchmarks](https://jer-k.github.io/effect-prophet/benchmarks/uncertainty.md)
- [Full Bayesian fitting benchmarks](https://jer-k.github.io/effect-prophet/benchmarks/bayesian.md)
- [Cross-validation and choosing settings benchmarks](https://jer-k.github.io/effect-prophet/benchmarks/cross-validation.md)