MMM Master Classes Season 2

Data processing is one of the most important steps in the marketing mix workflow. This stems from the variables that need to be transformed before being ingested into the model. In this series of videos, we explore several processors that allow the analyst to get data ready for the model. 

Data Transofrmation

MassTer Processors – Media Transformation in MMM

Episode 1

Decay/Adstock describes the percentage of the message that is lost period on period.

Diminishing Returns depicts the fact that as spend increases, the sales generated will increase as well, but at a decreasing pace until it reaches full saturation.

data smoothing

MassTer Processors – Data Smoothing in MMM 

Episode 2

Moving Average: It smoothens the input variable by sliding a window and calculating the local average using that window.

Median Filter: replaces each data entry with the median of neighboring entries. The pattern of neighbors is called the “window”, which slides, entry by entry, over the entire variable.

Seasonality: Detecting a seasonal pattern in sales is very important when dealing with highly seasonal products/ packages

Data Splitting

MassTer Processors – Data Splitting in MMM 

Episode 3

Splitter: mainly used to split data into multiple datasets. It allows the user to measure different impacts for different executions for the same media channel.

Split Region: generally used in Pooled Regression When the effect of a variable is expected to be different (lower/higher) than the regional average.

The Edit processor: enables the user to change some of the values of the raw data. They can also use it to correct some data entry anomalies. 

Data Splitting

MassTer Processors – Calendar Variables in MMM

Episode 4

Splitter: mainly used to split data into multiple datasets. It allows the user to measure different impacts for different executions for the same media channel.

Split Region: generally used in Pooled Regression When the effect of a variable is expected to be different (lower/higher) than the regional average.

The Edit processor: enables the user to change some of the values of the raw data. They can also use it to correct some data entry anomalies. 

More episodes coming soon!