For an e-commerce company, calculating effects of TV advertisement on Visits of Website and app:
We can follow easily our social media/google ads etc of effects on our website / app visit count or sales rate. But there is a limitation of the model is that offline marketing campaigns (e.g., TV, radio or billboard advertising) cannot easily be included (because we do not know which sessions were triggered by offline media). However, we have spotted that directly after a TV spot is aired more customers than we would expect to visit our online shop. We decided to use this signal to improve our attribution model.
In this notebook I want to take a detailed look into the data sets, each column's distributions, relationships between data sets and so on. You can start to read my home assignment from this step. Click here to go to EDA.ipynb
In this notebook I want to clean the data sets, create new columns to analyse tv spot interactions and calculate TV probabilities. This step is the second step of my home assignment.
Click here to go to TV_Probability.ipynb
You can find the helper functions that were used in Jupyter Notebooks. Click here to go to util.py