Amalie Updated by Amalie

  1. Raw log-level data
    Each pageview fires one (or more) pixel(s) that generates a single log line. These log lines contain date, time, IP, User Agent, Cookie identifier, and more. Raw log-level data refers to these data.
  2. Deep learning
    A method of machine learning that relies on automatic pattern recognition based on large amounts of raw log-level data. This is the base engine of the FIAM 2.0 reporting.
  3. Device reach
    The unique number of devices reached by the measurement. A single person can own multiple devices.
  4. Device type information
    Information extracted from the User Agent identifying is the device is a mobile, desktop, etc.
  5. Geographical information
    IP addresses carry information regarding location of origin of traffic.
  6. Census data
    Information pulled from the population census, either from Eurostat or Statistics Finland.
  7. Panel data
    Observed panelists contain information on age, gender, education, income, etc.
  8. Probabilistic modeling
    A method of distributing devices to human based on a synthetic population approach.
  9. Human reach
    The number of unique humans reached by a campaign
  10. Statistical extrapolation
    Up-scaling reach and sessions by the consent rate of raw log-level data, and downscaling to target group panelist ratio
  11. Profiled reach
    Total human reach distributed onto target groups

How did we do?