# approximate Bayesian computation

> computational method used to estimate the posterior distributions of model parameters

**Wikidata**: [Q21045424](https://www.wikidata.org/wiki/Q21045424)  
**Wikipedia**: [English](https://en.wikipedia.org/wiki/Approximate_Bayesian_computation)  
**Source**: https://4ort.xyz/entity/approximate-bayesian-computation

## Summary  
Approximate Bayesian computation (ABC) is a computational method used to estimate the posterior distributions of model parameters when direct likelihood calculations are infeasible. It provides a way to perform Bayesian inference for complex models, particularly in fields like population genetics and systems biology. ABC approximates the posterior by simulating data and comparing it to observed data.

## Key Facts  
- Short name: ABC  
- Also known as: ABC  
- Named after: Thomas Bayes  
- Subclass of: Bioinformatics  
- Freebase ID: /m/02rw1k1  
- Microsoft Academic ID (discontinued): 2779377595  
- Wikidata description: Computational method used to estimate the posterior distributions of model parameters  
- Wikipedia title: Approximate Bayesian computation  
- Available in Wikipedia languages: Catalan, German, English, Ukrainian  
- Sitelink count (Wikidata): 4  

## FAQs  
### Q: What is approximate Bayesian computation used for?  
A: ABC is used to estimate posterior distributions in Bayesian inference when calculating the likelihood function directly is computationally impractical. It is especially useful in complex models in fields such as population genetics, ecology, and systems biology.

### Q: How does approximate Bayesian computation work?  
A: ABC works by generating simulated datasets using parameter values sampled from a prior distribution. These simulations are then compared to the observed data using a distance metric. Parameters producing simulations close enough to the observed data (within a tolerance level) are retained to form an approximation of the posterior distribution.

### Q: Why is ABC needed instead of traditional Bayesian methods?  
A: Traditional Bayesian methods require computing the likelihood function, which can be impossible or computationally prohibitive for complex models. ABC bypasses this requirement by relying on simulations, making it applicable to a broader range of problems.

## Why It Matters  
Approximate Bayesian computation addresses a critical limitation in statistical modeling—how to perform Bayesian inference when the likelihood function is intractable. In many scientific domains, especially those involving complex stochastic processes, writing down the likelihood explicitly is not feasible. ABC enables researchers to fit realistic models to data without needing to compute likelihoods directly. This has had a transformative effect in areas like population genetics, where evolutionary processes are modeled as complex simulations. By enabling approximate inference in these settings, ABC expands the scope of probabilistic modeling and supports more accurate scientific conclusions in data-rich but likelihood-poor environments.

## Notable For  
- Enables Bayesian inference without requiring explicit likelihood calculations  
- Widely adopted in bioinformatics and computational biology, especially in population genetics  
- Relies on simulation-based approaches rather than analytical solutions  
- Offers flexibility in modeling complex, stochastic systems  
- Provides a practical workaround for high-dimensional and intractable statistical models  

## Body  

### Overview  
Approximate Bayesian computation (ABC) is a class of computational techniques used in Bayesian statistics to estimate posterior distributions when evaluating the likelihood function is computationally prohibitive or analytically intractable. Instead of computing likelihoods directly, ABC uses simulations from the model to approximate the posterior.

### Methodology  
The core idea behind ABC involves the following steps:  
- Sample parameter values from a prior distribution  
- Simulate data using these parameters  
- Compare the simulated data with the observed data using a distance measure  
- Accept the parameters if the distance is below a certain threshold  
- Repeat to build an empirical approximation of the posterior distribution  

This approach avoids explicit likelihood evaluations, making it suitable for complex models often found in biological and ecological sciences.

### Applications  
ABC is particularly prominent in:  
- **Population genetics**: Modeling evolutionary processes and demographic history  
- **Systems biology**: Inferring parameters in biochemical networks  
- **Ecology**: Analyzing species interactions and dispersal dynamics  

Its utility stems from its compatibility with simulation-based models where traditional statistical tools fail.

### Relationship to Bioinformatics  
As a subclass of bioinformatics, ABC intersects with computational biology and statistical genetics. It supports the analysis of large-scale biological data by offering robust parameter estimation tools for mechanistic models.

### Technical Characteristics  
- Does not require likelihood functions  
- Computationally intensive due to reliance on repeated simulations  
- Accuracy depends on choice of summary statistics and tolerance levels  
- Can be enhanced with machine learning techniques for dimensionality reduction  

ABC thus represents a powerful alternative to classical Bayesian inference in modern computational science.

## References

1. [OpenAlex](https://docs.openalex.org/download-snapshot/snapshot-data-format)