Plate notation is a useful visual method for describing graphical models, but the software can be awkward. Here we demonstrate daft-pgm, a solution using pure Python.
Posts in technical-series
In the final post of this technical series on derivatives trading we discuss using options as insurance, why the Black-Scholes model is wrong and trader psychology.
In the third post in this series on volatility and options trading we investigate the effects and trading implications of input behaviours on the option price.
Continuing our series on options and options trading, we focus on the behavioural patterns associated with options prices and how non-linear behaviour is an important consideration.
In this series we cover options: a deceptively complex trading instrument that provide an entirely different type of insurance - against directional moves in financial markets.
In the final article of this technical series we demonstrate hierarchical linear regression using PyMC3 to compare vehicle NOx emissions for a range of car manufacturers.
In the second article of this technical series we demonstrate the flexible syntax of PyMC3 with regularized linear modelling of car emissions data and model evaluation.
Bayesian inference bridges the gap between white-box model introspection and black-box predictive performance. This technical series describes some methods using PyMC3, an inferential framework in Python.
We finish our series on Bayesian networks by discussing conditional probability, more complex models, missing data and other real-world issues in their application to insurance modelling.
We continue our series on Bayesian networks by discussing their suitability for fraud detection in complex processes: for example assessing medical non-disclosure in life insurance applications.
Bayesian networks are useful tools for probabilistically computing the interdependencies and outcomes of real-world systems given limited information. Here we describe their use in fraud detection.
Practical data science projects often include an aspect of anonymisation to carefully remove sensitive information prior to analysis; here we demonstrate several complimentary techniques and principles.
In this technical article we explain why and how to use Singular Value Decomposition (SVD) for feature reduction: making large datasets more compact whilst preserving information.
Real data is often unavailable for creating demos, learning and especially for publishing. Here we describe methods to generate realistic artificial data which has fewer constraints.
In the final article of this technical series we generate outputs to estimate the value of mortality swaps and discuss the concept of pricing tail risk.
In the second article of this technical series we put the pieces together to begin estimating the value of a mortality swap, an esoteric insurance derivative.