Analysis of Covid-19 lockdown effects on pollutant concentration levels in Danmarks Plass measuring station in Bergen, Norway.

Pulling data from their respective APIs

• Vehicle volume data from https://www.vegvesen.no/trafikkdata/api/
• Air quality data from https://api.nilu.no/
• Meteorological data from https://frost.met.no/ (needs personal API key https://frost.met.no/auth/requestCredentials.html)

Cleaning, tidying and then combining all the datasets by Date (hourly) column.

• Traffic data has a "Coverage" metric that gives a rating for data quality. Under 95% coverage hours will be omitted from research as it would undercount the number of vehicles. https://www.vegvesen.no/trafikkdata/start/om-trafikkdata
• Adding seasonal variables and seasonal dummy variables

Creating comparable periods between Lockdown months in 2020 and same weeks/days from previous years of 2017, 2018 and 2019.

• Omitting holidays as they would have unusually low traffic flow. Omitting weekends due to change in commuting patterns.
• Transforming negative values of pollutant concentrations to absolute value over 2.
• Inspecting the distribution of NAs in pollutant concentrations.

Summary statistics of hourly data on selected time periods

• Comparison of Median, SD, IQR, Min, Max, and Coefficient of Variation on pollutant concentration levels and vehicle volume numbers.

Visualizations of the data

• Average hourly pollutant concentration between periods.
• Mean hourly distribution of vehicle length volumes and pollutant concentration levels.
• Average daily pollutant concentration comparison between lockdown and pre-lockdown periods.
• Average daily vehicle volume comparison between lockdown and pre-lockdown periods.

Statistical tests

• Linear models of the data with different variables as input.
• Multicollinearity and autocorrelation present.