As a project for a course in Machine Learning applications on time series analysis, I created some functions to work with time series data (different frequencies) coming from the Central Reserve Bank of Peru (BCRP). The first function helps to arrange a dataframe with several variates of the same frequency given their codes, the desired time period and their frequency. The second function plots the very variables extracted with the first function employing STL decomposition (it allows to choose between the additive and multiplicative model). Then, we propose three functions that obtain the best ARIMA, ARMA and ETS model for a given time series variate (we constructed three different functions in order to save computational resources) employing crossvalidation. Finally, we built a forecasting function that employ bootstrapping to build confidence intervals for each prediction. All these functions are accompanied by several applications and examples. Every comment on this code is in Spanish and every possible mistake is only responsibility of the author.
Joselu14/Univariate_time_series_analysis_BCRP_data
As a project for a course in Machine Learning applications on time series analysis, I created some functions to work with time series data (different frequencies) coming from the Central Reserve Bank of Peru (BCRP).
Jupyter NotebookGPL-3.0