gasoline-time-series-analysis

Gasoline’s seasonal price prediction in Italy 3 months in advance prediction of the price of gasoline for final consumer (EUR) based on monthly data.

DATASET

We create our own dataset combining multiples sources:

• Monthly employment rate in Italy (2004-today) (ISTAT)
• Yearly vehicle gas emissions (1990-today) (ISPRA)
• Monthly oil’s price in Italy (1996-today) (IndexMundi)
• Daily EUR-USD exchange rate (2004-today) (YAHOO Finance)
• Daily ENI stock price (1996-today) (YAHOO Finance)

The evaluation of the models identify only some of them to be relevant. What follows take only them into account.

Merge the data

Data comes from 3 different data sources, contained in three files: prezzi_mensili_benzina_dal_1996_a_20221028.csv (MISE), oil_price_monthy.xlsx (IndexMundi) and DatiCopertTrasportoStrada1990-2020.xlsx (ISPRA), employement_rate.csv (ISTAT).

Interesting information is extracted. To obtain the merged dataset run: $ Rscript prepare_data.R infer_NAs_bool

infer_NAs_bool = true if you want to generate also the dataset without NAs. They are infered using a polynomial model.

Generate plots

To save the plots in the /plotsdirectory run

$ Rscript plots.R