Skip to content
Paul Newnes edited this page Jul 24, 2019 · 5 revisions

This R package is a wrapper to the channelAttribution package.

The markovreport function takes a data frame of clickstream data and produces a report using non-heuristic (Markov) and heuristic based attribution models.

You can extract clickstream data using a service such as Scitylana .

Pre-requisites are:

  • The source / medium channel has been cleaned / consolidated into a new column. It's recommended you do this to prevent different iterations of the same source being counted separately.
  • There is a financial value assigned to each conversion in the column M_transactionRevenue. This could be automatically via Google Analytics goal or ecommerce tracking, or manually. If there is no value assigned then simply duplicate the conversions columns to a new column called M_transactionRevenue.

The format is:

  • markovreport(data, date_from, date_to, look_back, channels, goal_number)
  • markovreport(mydataframe, "2019-06-01", "2019-06-30", 30, "my_cleaned_source_variable", 1)

Parameters:

  • data is the name of the clickstream data frame
  • date_from and date_to parameters specify the period for transactions
  • look_back is the window to consider visits before the specified transaction period
  • channels is the variable within the specified data frame that contains the cleaned source channel
  • goal_number is the Google Analytics goal number that represents a purchase
Clone this wiki locally