Addressing challenges in different business areas using tried-and-tested econometric models: fraud detection, outflow forecast, marketing modeling, etc. We use Python, Deep Learning frameworks, NoSQL & Big Data, BI and other technologies.
EXPERTISE IN DATA SCIENCE
- Customer clustering and segmentation
- Demand forecasting and marketing campaign optimization
- Prevention of customer outflow
- Loan scoring for banks
- Optimization of production plan and resources
- Semantic analysis of feedback
The IBA Group team has created 20+ Data Science scenarios for banks, industrial, energy, government-owned and other companies. Contact us and we will send you details.
Python: pandas, numpy, scipy, sklearn, matplotlib, nltk, catboost, xgboost, gensim, spacy
Deep Learning frameworks: TensorFlow, OpenCV, PyTorch, Keras
Web: Flask, Node.js, Django, HTML5, CSS
NoSQL & Big Data: PySpark, MongoDB, Elasticsearch
BI: Cognos, Tableau, Kibana
Other: R, Docker, SQL
1/ CUSTOMER CLUSTERING AND SEGMENTATION
Develop a customer clustering model:
- what is a cluster with the best product attachment rate
- which cluster is the least effective for cooperation
- what is the profile of the most successful customer for cooperation in the context of geographical spread
- Statistical analysis and data origination
- Building clustering models
- Data visualization with BI tools
- Development of reports and presentations based on survey results.
2/ DEMAND FORECASTING AND MARKETING CAMPAIGN OPTIMIZATION
Analysis and forecasting of demand for goods.
Building an optimal advertising strategy.
- Creation of an interpreted forecast model with a seasonal component.
- Adjustment for response curve and the attenuation effect.
- Calculation of the optimal advertising company for various scenarios.
- Determination of the halo effect.
Infopanel for benchmarking analysis of sales and advertising performance in various channels.
Calculator of the optimal budget allocation for the forecast period.
Python, Tableau, MS Excel.
3/ PREVENTION OF CUSTOMER OUTFLOW
Prediction of customer outflow.
Analysis of characteristics affecting outflow.
- Collection, cleaning and origination of data.
- Analysis of patterns and determination of their significance.
- Creating metrics to measure customer behavior.
- Creating a model that predicts customer outflow.
- Implementation of the solution.
Prediction of customers prone to breach of contract.
Identification of key indicators affecting customer outflow.
Python, SQL, DB2, Linux, bash.
4/ LOAN SCORING FOR BANKS
Reduce credit risks and bank losses arising from bad loans; increase the profitability of the retail business.
- Aggregation of customer information from various sources.
- Identification of key factors affecting the likelihood of an event.
- Development of models that predict the likelihood of an event.
- Interpretation of model results.
The solution helps automate the credit risk assessment process, generate scoring cards, and reduce the human factor in the application process.
Application scoring: assessing the creditworthiness of borrowers to obtain a loan.
Behavioral scoring: assessment of likelihood of repayment of loans already issued.
Collection scoring: assessment of likelihood of full or partial repayment of a loan by the borrower in case of delay in debt repayment.
Fraud scoring: identification and prevention of fraudulent actions by potential and existing borrowing customers.
5/ OPTIMIZATION OF THE PRODUCTION PLAN AND RESOURCES
Modeling the optimal production plan.
Selection of scenarios and modeling parameters.
- Implementation of download and maintenance of sales plans, standards and resource restrictions.
- Building an optimization model.
- Calculation of the achieved performance indicators and visualization of the optimal plan using BI tools.
An interactive tool has been created that allows modeling an optimal production plan with predetermined resource restrictions.
IBM ILOG CPLEX, IBM Cognos BI.
6/ SEMANTIC ANALYSIS OF FEEDBACK
Create a system for analyzing the activities of banks based on collected customer feedback.
- Collection of feedback from public websites.
- Creating a sentiment analysis system.
- Creating an information panel.
- Benchmarking analysis of banks.
- Improved efficiency and quality of business processes in the analyzed bank.
- Better efficiency of decision-making on improving the bank’s operations.
Python, MySQL, Qlik Sense.