I want to analyze fuel prices in germany and investigate in some forecasting models to predict the fuel price in the future. Therefore I used the history data available from Tankerkoenig, they are available in the azure cloud for private use with the following license https://creativecommons.org/licenses/by-nc-sa/4.0/.

History Data from Tankerkoenig History if you start in with Data Reading.

• pandas
• glob
• math
• matplotlib.pyplot
• plotly
• numpy
• sklearn
• keras
• sqlalchemy
• statsmodels
• pylab

If you downloaded the History Data (2015-2019) from Tankerkoenig you can start with DataRead-Part. And have a look at the Data Structure section below.

For better handling i have created a slice out off the data, one year of data has around 5 GB. So i collected only stations with PLZ beginning with 40. Loaded the data per year and saved the data for the few stations in a sql file for later use. The needed functions for loading and saving data and for getting a dataset for a special station i've put in helper_functions.

Therefore you can start directly with DataPreparation there i've done some data analysis and graphical plottings with this data. There i read the data from a sqllite database created before prices_40.sql. There i've put 2 tables inside Prices and Stations.

For the forecasting i've created an extra jupyter notebook Forecasting.

You'll find a discussion of this Project in [Forecasting of Fuel prices in germany](./Forecasting of Fuel prices in germany.pdf)

I've put the data in one directory "sprit" and inside this directory there where folder for every year and month. Inside the folders there is a csv-file for every day. So we have ./sprit/year/month/year-month-day-prices.csv for example ./sprit/2015/12/2015-12-01-prices.csv. That is the same structure as it is on Tankerkoenig, despite of the stations.csv.

sprit
+-- stations.csv
+-- 2015
| +-- 01
| | +-- 2015-12-01-prices.csv
| | +-- ...
| +-- 02
| +-- ...
| +-- 12
+-- ...
+-- 2019
| +-- 01
| +-- 02
| +-- ...
| +-- 12

Inside the csv the data is structured as follows:

date,station_uuid,diesel,e5,e10,dieselchange,e5change,e10change

The corresponding stations are in the ./sprit/stations.csv file and have the following structure inside:

uuid,name,brand,street,house_number,post_code,city,latitude,longitude

They are connected via the UUID in both files.

def get_data4uid(uid, typ):
    """
    Give the typ data for given uid
    Input: uid, typ
    Output: Dataframe"""
    data_uid = pd.DataFrame(prices_pd[prices_pd['station_uuid'] == uid][['date', typ]])
    
    # has to be changed, but for now utc=True
    data_uid.date = pd.to_datetime(data_uid.date, utc=True)
    
    data_uid.set_index('date', inplace=True)
   
    
    return data_uid