Track 1: Datarathon
VIB wants to come up with an innovation for making use of their existing data to improve the mobile banking customer experience. VIB is excited about the out-of-the-box ideas that can determine factors that make our customers leaving the MyVIB mobile application.
Provide a Word/PDF document in which clearly demonstrates:
Definition of customers who leave MyVIB application based on given dataset. Brief proposal describing some Data Preprocessing strategies (involving Feature engineering, if any) and Functionality in Data Mining (E.g. Regression, Classification, Clustering, so forth). Give appropriate reasons. Download data here
- Understanding business context in a correct way.
- Give a clear and appropriate definition of customers who leave MyVIB application based on our dataset.
- Appropriately perform exploratory data analysis with best Feature engineering.
- A brief model proposal which describes which functionality in data mining (e.g. regression, classification, clustering, and so forth) you use.
- 1 PPT for visualization presentation.
- 1 Word/PDF for detailed data mining process, including Exploratory Data Analysis, Data Preprocessing, Data Mining methodologies and Performance Evaluation (Give appropriate reasons for your choices). All your code script files (.py, .R).
- A comprehensive documentation which expresses your data mining process in detail, including the algorithm you employ.
- An informative visualization which can be deliverable to persons who are not expertised in data science.
- Criteria: determine factors and quantify their effect, depending the model and functionality you use. For example, if you use classification algorithms, we will base on Accuracy and Recall ratios. If you use regression algorithms, we will check the model prediction performance.
We will rank in the descending order the models based on their appliability, interpretability, and performance.