Accenture: Optimize campaign targeting via LightGBM with native categorical encoding. Python, LightGBM, Scikit-learn, Pandas.
You are working for Accenture as a Data Scientist. Accenture has been commissioned by an insurance company to develop a tool to optimize their marketing efforts. They have given us a data set as a result of an email marketing campaign. The data set includes customer information, described below, as a well as whether the customer responded to the marketing campaign or not.
Design a model that will be able to predict whether a customer will respond to the marketing campaign based on his/her information. In other words, predict the responded target variable described above based on all the input variables provided.
Briefly answer the following questions:
Files:
marketing_training.csv - contains the training set that you will use to build the
model. The target variable is responded.marketing_test.csv – contains testing data where the input variables are provided
but not the responded target column.Descriptions of each column:
| Type | Name | Description |
|---|---|---|
| Input Variables | custAge | The age of the customer (in years) |
| Input Variables | profession | Type of job |
| Input Variables | marital | Marital status |
| Input Variables | schooling | Education level |
| Input Variables | default | Has a previous defaulted account? |
| Input Variables | housing | Has a housing loan? |
| Input Variables | contact | Preferred contact type |
| Input Variables | month | Last contact month |
| Input Variables | day_of_week | Last contact day of week |
| Input Variables | campaign | Number of times the customer was contacted |
| Input Variables | pdays | Number of days that passed by after the client was last contacted from a previous campaign (numeric; 999 means client was not previous contacted) |
| Input Variables | previous | Number of contacts performed before this campaign for this client |
| Input Variables | poutcome | Outcome of the previous marketing campaign |
| Input Variables | emp.var.rate | Employment variation rate - quartlerly indicator |
| Input Variables | cons.price.idx | Consumer price index - monthly indicator |
| Input Variables | cons.conf.idx | Consumer confidence index - monthly indicator |
| Input Variables | euribor3m | Euribor 3 months rate - daily indicator |
| Input Variables | nr.employed | Number of employees - quarterly indicator |
| Input Variables | pmonths | Number of months that passed by after the client was last contacted from a previous campaign (numeric; 999 means client was not previous contacted) |
| Input Variables | pastEmail | Number of previous emails sent to this user |
| Target Variables | responded | Did the customer respond to the marketing campaign and purchase a policy? |
Provide the following:
.csv file containing the predictions of the test data. You can add the target column (responded) to the test data or simply provide it alone with the id column.