Use dataclasses in your FastAPI projects, and speed up your pydantic model operations — Python dataclasses are fantastic. Pydantic is fantastic. It is a tough choice if indeed we are confronted with choosing one or the other. I would say that comparing these two great modules is like comparing pears with apples, albeit similar in some regards, different overall. Pydantic’s arena is data parsing…

Simple cointegration methods in Python. The Engle-Granger approach, the most intuitive method. — The mathematics and concepts of most cointegration methods are not always very straightforward. Often, complex mathematical tools obscure the intent and steps of the methods. Such complexities also detract some from further exploration of the subject. The Engle-Granger approach to cointegration does not suffer from this. It may not be…

Class interfaces, abstraction layers, inheritance, isn’t that a software developer problem? Why should you, as a data scientist, care? — Interfaces make almost all of our favorite data science libraries possible. That is a good enough reason, at least for me, to care. But let us go deep into the subject. …

Never mind the maneuvers; minimize the Dickey-Fuller statistic directly to achieve a stationary time series. Full Python code. — There are many methods for finding cointegrating vectors. Some methods leverage OLS, others matrix eigendecomposition. Regardless of the method, it is customary to do a sanity check by performing a stationarity test on the resulting time series. The workflow goes something like this:

Part 5 of the Stochastic Processes Simulation series. Simulate generalized Brownian motion in Python. — Alas, we come to the end of our journey through the Stochastic Processes Simulation series. But we saved the best for last. In this story, we will explore generalized geometric Brownian motion. Generalized geometric Brownian motion occurs when the quotient of the process differential, and the process itself follows an…

Part 4 of the Stochastic Processes Simulation series. Simulating geometric Brownian motion in Python from scratch. — Geometric Brownian motion is perhaps the most famous stochastic process aside from Brownian motion itself. It arises when we consider a process whose increments’ variance is proportional to the value of the process. In this story, we will discuss geometric (exponential) Brownian motion. We will learn how to simulate such…

Part 3 of the Stochastic Processes Simulation series. Simulating the Cox-Ingersoll-Ross process in Python from scratch. — The Cox-Ingersoll-Ross (CIR) process is another stationary process of great importance. Albeit most of its applications lie in the financial realm, it is an interesting stationary stochastic process in its own right. It is the simplest particular case of basic affine jump diffusive processes and an ergodic process.

Class, dataclass, attrs, pydantic or NamedTuple, which container is the most suitable for your project? Let’s find out. — We start a greenfield project, motivation is high. After the basic planning is done we begin the initial typing of our app to be, we are about to build the data models that will flow through our app’s pipelines. …

Part 2 of the Stochastic Processes Simulation series. Simulating the Ornstein-Uhlenbeck process, the canonical stationary process, in Python. — The quest for stationarity. Doesn’t matter if we want to perform statistical inference or fit a machine learning model, we always seem to be searching for stationary features, stationary explanatory variables and well, stationary everything. When we deal with indexed random variables stationarity means stationary stochastic processes.