Cuando se trabaja con datos de series temporales, es importante evaluar la calidad del modelo de forma que refleje con precisión las situaciones del mundo real. Generalmente, el proceso simplificado de construcción de un modelo de Machine Learning es el siguiente: Esta es la forma en que se hace tradicionalmente Read more…
ARCH/GARCH models can assist or improve the forecast of ARIMA models. Firstly, time series data must be stationary for these models to be applicable. If not, we should apply some transformations such as differencing. Since these models are mainly used in finance we will refer to the price returns from Read more…
Time series can be seasonal. What does this mean? It means that our data exhibits a season or cycle that regularly repeats, for example, weekly, monthly or annually. There is something important to note. The presence of a cycle structure does not mean seasonality. For this cycle to be seasonal Read more…
ARIMA models are a simple but powerful way of modelling time series. They were introduced in the previous article, however, their parameters must be selected correctly for them to offer accurate forecasting. An ARIMA(p,d,q) model is defined by three parameters: p : autoregressive order d : difference order q : Read more…
A time series is a series of data points ordered in time. Time is often the independent variable and the objective is usually to make a forecast for the future. It has many applications, some of which are forecasting sales, stock prices, temperatures or even the number of COVID cases! Read more…