Fig 1.
Natural logs of hourly ADAC traffic jam reports (top left), its autocorellations (top right), spectral density (bottom left) and histogram of percentage changes (bottom right). Data Source: ADAC (adac.de) and own calculations.
Fig 2.
The reduced series are searches containing the word stau without containing any of: nrw, a7, a2, a3, a1, wdr, a8, a5, aktuell, autobahn, a9, info, swr3, bayern, hamburg, a4, staumelder, adac, berlin, a6, verkehr, ffh, hessen, köln, münchen, swr, a81, a61, deutschland. A 60% of the total stau search volume is accounted for by these words. Data Source: Google Trends (www.google.com/trends) and own calculations.
Table 1.
Origins of four highway-specific Google “stau” searches.
Fig 3.
Hourly Google Searches for “stau.”
The natural logs of the hourly stau search intensity (top left), its autocorellations (top right), spectral density (bottom left) and histogram of percentage changes (bottom right). Data Source: Google Trends (www.google.com/trends) and own calculations.
Fig 4.
Google Searches for “stau” vs ADAC reports.
A cross correlogram between the hourly number of ADAC traffic jam reports and the hourly Google search intensity for stau, based on the entire observation time interval of 51 days, establishes that Google search has a two hour advance on road conditions. Data Source: Google Trends, ADAC and own calculations.
Table 2.
Forecasting the number of ADAC traffic reports.
Fig 5.
Clearly the regression line of Eq M5 captures the data better than that of Eqs M4 and M3. A higher degree term is apparent (bottom right). Data Source: Google Trends (www.google.com/trends), ADAC and own calculations.