Motivation for launching the project by the customer:
the need to increase the involvement of bank clients in the use of internal products revealed limitations in the current solution: segmentation and profiling models were of low quality, the identified user segments were not interpretable, analysts could not use the modeling results.
Description of the initial situation:
Project goals:
creating a model for profiling and segmenting the customer base by their transactional activity with a high level of quality and interpretation.
MIL Team solution:
the use of the team’s existing solutions in the field of customer analytics and transactional data analysis made it possible to implement a procedure for training temporal models for segmenting the customer base.
To build the model we used:
Simulation results:
Customer: Finance, Banking
Technology stack: TopicNet, BigARTM, gensim, Python
the need to increase the involvement of bank clients in the use of internal products revealed limitations in the current solution: segmentation and profiling models were of low quality, the identified user segments were not interpretable, analysts could not use the modeling results.
Description of the initial situation:
- Segmentation of the bank's client base assumes high conversion of internal product offerings to clients;
- Segmentation requires deep understanding and analysis of bank customer behavior, such as user transaction data.
- An increase in the quality of segmentation leads to an increase in response to the bank’s products offered and affects business performance: from customer loyalty to profit.
- The segmentation models used were not interpretable and their quality was low;
- For this reason, there were significant risks when using current models in product functionality.
Project goals:
creating a model for profiling and segmenting the customer base by their transactional activity with a high level of quality and interpretation.
MIL Team solution:
the use of the team’s existing solutions in the field of customer analytics and transactional data analysis made it possible to implement a procedure for training temporal models for segmenting the customer base.
To build the model we used:
- Bank customer transactions;
- Description of MCC codes and merchants;
- Accompanying information about the user;
- Results of sales of banking products.
Simulation results:
- Behavior model of bank users;
- Model for predicting the likelihood of additional purchase of a banking product;
- Segments of the customer base.
Customer: Finance, Banking
Technology stack: TopicNet, BigARTM, gensim, Python
