Madriss Seksaoui - Email marketing professional thesis - Msc in Data Sciences and Business Analytics
This git repository contains all the necessary script files and anonymized sample data sets to reproduce all the results from:
Three sample data sets are provided:
- Fahsion : sample_campaign_fashion.csv
- High tech : sample_campaign_high_tech.csv
- Retail : sample_campaign_retail.csv
Note that each sample data set contains 10 000 instances, the open proportion and other statistics are given in the table below:
| Dataset | Open proportion |
|---|---|
| Fashion | 8.26 % |
| High tech | 9.69 % |
| Retail | 9.64 % |
File description:
- train_test_vizu.R : Provides the user with the plots from Chapter 3 to illustrate overfitting/underfitting
- Preliminary_Modeling.R : Trains benchmark models including CART, RF and Logistic Regression
- model_comparison.R : Trains and compares the models listed in Chapter 5, the ensemble model is also trained in this file
- Cleaning_Functions.R: Contains several functions to obtains explanatory variables from the sample data sets
- get_text_features.R: The file contains a set of functions to extract features from subject lines
- Email_Campaigns_EDA.R: This file contains the code for the plots appearing in Chapter 4
- Email_Campaigns_EDA_Adv.R: This file is used to obtain more advanced plots
- Text_Analysis.R: Is used to to train models using the BOW and 2-Gram features from the subject lines