Titre : Mathematical optimization of time series. Cleaning financial series of calendar and seasonal effects ‐Moroccan financial market‐ 

Auteurs : Mustapha Bouchekourte, Norelislam El Hami,  

Revue : Uncertainties and Reliability of Multiphysical Systems 

Numéro : Issue 1 

Volume : 2 

Date : 2018/05/16 

DOI : 10.21494/ISTE.OP.2018.0260 

ISSN : 2514-569X 

Résumé : The construction of the national calendar allows the optimization and the production of the seasonally adjusted series, thus significantly improving the quality of the models used to obtain reliable, legible, and interpretable indicators. Series decomposition improves the quality of regressions by isolating irregular, seasonal, and time-sensitive effects that can bias the modeled relationships. We will analyze the series of variables explored using the Demetra + software [1] to better refine our estimates. The exploratory analysis of the data used should make it possible to
summarize the distribution of each series and the relationships between variables whose characteristics might require measurement and unit transformations (or recoding). Outliers, for example, are likely to influence the results of a statistical model. This treatment will capture the influential values before modeling the correlations between the variables being estimated. 

Éditeur : ISTE OpenScience