case study
Machine learning case study for an Insurance company to detect fraud
Challenge, Context, Problems to be solved
Fraud detection in the commercial network
Mission, tools and methodology
Fraud detection: internal, of customers, of suppliers, associated branches ...
Fraud patterns were detected optimizing the cost with respect to the income for each unit studied.
Achieved results
Reduction of costs derived from fraud.
case study
Machine learning use case for a Debt collection software powered by AI
Challenge, Context, Problems to be solved
Detection of customer behaviors to achieve greater recovery.
Mission, tools and methodology
Collections optimization with ML Python
The key variables to increase the recovery rate in clients were detected. Strengthening and improving them thanks to a strategy of continuous improvement on the results of the analytical engine.
Customers have also been segmented according to behavior. Own data has been integrated with 3rd party data. A BI reporting has been created with Tableau to improve internal knowledge and give more value of the product to the end customer.
Achieved results
Recovery KPI increase
case study
Machine learning use case for a leading provider of Information solutions on users behaviour
Challenge, Context, Problems to be solved
Detection of buying patterns and recommenders.
Mission, tools and methodology
Buying patterns of certain books, groups ... were detected to be applied in all countries through the commercial network. In addition to an assistant for the commercial network to understand what customers needed according to their profile.
Achieved results
Understand customers to offer just what they need.
case study
Machine learning case study for an AdTech company to detect fraud
Challenge, Context, Problems to be solved
Detection of fraud in online ads
Mission, tools and methodology
Customers were found to be using ads fraudulently to earn revenue from their apps.
Operational and financial data were cross-checked to quickly detect who was committing fraud and what patterns they were using. Used Redshift, Python & Knime.
Achieved results
Reduction of costs derived from fraud.
case study
Machine learning use case for a chain of Supermarkets to increase customer loyalty
Challenge, Context, Problems to be solved
Increase customer loyalty
Mission, tools and methodology
Thanks to the purchase data, patterns were detected to generate offers that increased customer loyalty.
Achieved results
A double benefit is achieved, on one hand increasing customer loyalty ratios.
And on the other hand, increase sales thanks to the understanding of the customer, the detection of behavior patterns and the optimization of offers.
case study
Machine learning use case for a Sports center management company to reduce churn
Challenge, Context, Problems to be solved
Reduce the number of people cancelling their memberships
Mission, tools and methodology
Detect churn patterns, areas with churn increases, profiling, customer assessment, recommend actions
Achieved results
Reduction of churn.
Greater benefits are achieved by understanding the client, adapting the offer to the demand, understanding the client's life cycle, optimizing prices / offers, detecting problems in the low / high.
case study
Machine learning use case for an Airline company to manage revenue with Alteryx
Challenge, Context, Problems to be solved
Manage and optimize revenues.
Mission, tools and methodology
Optimize costs and increase the benefits derived from overbooking, planning, routes, etc.
Achieved results
Analysts were empowered to optimize costs and increase the benefits derived from overbooking, planning, routes, etc.
case study
Machine learning use case for a Broadcast media
Challenge, Context, Problems to be solved
Conversation detection
Mission, tools and methodology
Analyze conversations on different platforms and social networks about the contents of the CCMA to detect trends, ideas, news ...
Achieved results
Increased understanding of the audience.
case study
Machine learning use case for a furniture Retail
Challenge, Context, Problems to be solved
Marketing campaign optimization.
Mission, tools and methodology
Detection of customers around the stores and which offers worked best with what type of customers.
Achieved results
Higher return on investment in marketing.
case study
Machine learning use case for a Financial services company to reduce non payments
Challenge, Context, Problems to be solved
Avoid non payments and recover unpaids.
Mission, tools and methodology
Analyzed behavior patterns to detect when customers were to become outstanding. Detect what actions to take to recover unpaid clients.
Achieved results
Improvement in non payment rates and better customer service.
Must have Machine learning Skills
Machine learning engineers must have a complete set of capabilities and experience:
- Data analytics & machine learning strategic consulting.
- Good Business Analysis experience
- Development, management & implementation of business analytics projects.
- Capacity to choose the technologies to use and architecture to implement.
- Innovation based on data (products, services, etc).
- With a broad functional view and use case exposure: improve commercial offer, improve recruitment, reduce customer cancellations, help locate new stores, reduce costs, cross-selling, up-selling, fraud detection, etc.
Other skills needed:
- Business intelligence
- Business analytics
- Sales
- Partner relationship management
- Data science & machine learning
- Cloud & distribution computing
- Business development
- Relationship management
- Data engineering & DW
- Big data
The technical knowledge of a data analyst should include the command of various of the following solutions, frameworks and languages:
- Snowflake · AWS · Google Cloud/AI · MS Azure · IBM Watson · Oracle · Hadoop · SAS · Splunk · Kubernetes · SAP Hana · Elastic · Salesforce
- Qlikview · Tableau · Alteryx · Trifacta . Power BI · Google Analytics
- Python · Java · R · Spark · SQL · MQL
Responsibilities of this role
End-to-end responsibility in the implementation of the data analytics project.
Very versatile profile based on real needs in data analytics projects, where the key is to achieve an impact on the business aligned with the objectives and its strategy. This requires understanding the needs of the client (internal or external), how to apply the analytics of data to meet the defined goals and how to achieve this in the minimum time & cost and with 100% guarantees.
Challenges and hot topics for Machine learning engineers
The main cross-cutting challenges for all companies would be:
- Customer analytics (analysis of customer information): Market analysis, sales optimization, improved understanding / relationship with the customer and prediction of customer behavior (additions / deletions / changes ...).
- Operations Analytics: Analysis of the supply chain (supply, production, warehouse management, transport and distribution at the point of sale) and the new applications linked to the Internet of Things (IOT) and geographic information systems.
- People analytics (talent within the organization): Analysis for effective strategic management of human resources, so that business objectives can be met quickly and efficiently, obtaining optimal performance on human capital.