1
Introduction
1.1
Task Background
1.2
Goals and Requirements
1.3
Conventions and Notes
2
Data
2.1
General Structure
2.2
Exploratory Data Analysis
2.2.1
Data Types
2.2.2
Targets
2.2.3
Skewness
2.2.4
Correlations
2.2.5
Donation Patterns
3
Experimental Setup and Methods
3.1
Tools Used
3.2
Data Handling
3.3
Data Preprocessing
3.3.1
Cleaning
3.3.2
Feature Engineering
3.3.3
Imputation
3.3.4
Feature Selection
3.4
Prediction
3.4.1
Setup of the Two-Stage Prediction
3.4.2
Optimization of
\(\alpha^*\)
3.5
Model Evaluation and -Selection
3.5.1
Evaluation
3.5.2
Selection
3.5.3
Dealing With Imbalanced Data
3.5.4
Algorithms
4
Results and Discussion
4.1
Preprocessing With Package kdd98
4.2
Imputation
4.3
Feature Selection
4.4
Model Evaluation and Selection
4.4.1
Classifiers
4.4.2
Regressors
4.5
Prediction
4.5.1
Prediction of Donation Probability
4.5.2
Conditional Prediction of the Donation Amount
4.5.3
Profit Optimization
4.5.4
Final Prediction
5
Conclusions
References
Appendix
A
Software
A.1
Python Environment
A.2
Package kdd98
A.2.1
Usage
A.2.2
Installation
B
KDD Cup Documents
B.1
Cup Documentation
B.2
Data Set Dictionary
Profit maximization for direct marketing campaigns
4
Results and Discussion
Below, the results from data preprocessing, model evaluation/-selection and predictions are shown.