4 MJV Data Science cases
“Without data, you are just another person with an opinion.” This quote by Edward Deming is from the last century. However, it looks like it was yesterday. It is more evident than ever that the data-oriented mindset is the cornerstone for decision-making and innovation.
Even with the tools and techniques that we have available today, many organizations find it challenging to be guided by the data they have. Today, we brought 4 Data Science cases to inspire and show you the unlimited possibilities it can bring to your company.
1- Fraud propensity indicator
One of our clients in the insurance industry was struggling with finding the necessary information within their data visualization tools and not identifying cases prone to fraud or unnecessary examination requests in processes.
The idea was to analyze thousands of daily health insurance claims to identify behavioral patterns of fraud called a receipt break.
The big challenge was to gather, sort, structure, and visualize all the previously unstructured data that was not used to generate insights and improve decision-making.
MJV used the Oracle DataBase to store, structure, and cross-reference data. The idea was to act on two different fronts:
- We use the database to integrate the data collected from the refund systems. Thus, it was possible to store and process digital data that was varied and unstructured
- We organize Dashboards to view specific data related to fraud prevention.
With this tool, the insurer is now able to process approximately 10,000 reimbursement requests per day. This way, it automated the identification of claims that present the possibility of fraud.
Dashboards provide all the visualization necessary for decision-making in real-time. In addition, it is now possible to make a more accurate and focused analysis on cases that are prone to fraud.
→ Want to learn more about Data Science? Then download our Data Science ebook!
2- Rethinking the Marketing Execution Score
In this case, a company in the FMCG industry needed to analyze and use the data available to improve their business strategies.
The Marketing Execution Score assesses a store’s ability to implement sales and marketing strategies that have been assigned to it. It turns out that it was necessary to understand how the metric worked to identify patterns.
Stores with a high Marketing Execution Score did not always have high sales volumes. We had to synthesize, cluster, and analyze the available data to understand the parameters correlated with the sales volume.
Only by making this type of identification would it be possible to establish best practices and define where investments would occur in the future.
MJV used Python, Jupyter Notebook, and Sklearn as tools to analyze the data. Based on that, it clustered the stores and tested statistical hypotheses within the created clusters. We were able to deliver the behavior of the sales volume by the levels of execution.
In addition to clustering, MJV delivered performance analysis, statistical assessments by sales channels and provided suggestions for reassessing the impact of items on the overall score of stores.
Thus, we established new parameters for the client to consider when defining the scores. This was instrumental in improving the company’s visibility and deciding where the investments would take place.
3- Targeting models for pop-up stores
Another interesting case was the creation of segmentation models for pop-up stores. The client was in the insurance industry, and the project involved skills in areas such as Artificial Intelligence, GeoMarketing, Design Thinking, and, of course, Data Science.
The great challenge for MJV was to create a segmentation model based on data that could cross-check data from the customer base with the census so it would be possible to deliver the best experience and customization according to the location of the stores.
With that, it was expected to redefine both the sales space and the experience that brokers would promote, bringing high-quality leads. In addition, it allowed them to create hyper-customized journeys.
We took the database and used exploratory analysis to gain insights. After that, we used AI clustering models. With the support of APIs to map the latitude and longitude information, we updated the database. We integrated it with a visualization tool so that the customer had access to a data-driven platform, which allows segmentation by areas and according to the demographic of the population—a perfect combination of internal and external data sources.
We were able to redesign the sales experience, which improved the brand’s presence and led to a closer relationship between brokers and customers. In addition, it was possible to create a strategy to position the stores, identify audiences that make sense, and create personalized offers for each one.
4- Stock Exchange and Design Driven Data Science
Finally, we will present the case with one of the largest capital exchanges in the world. In this case, MJV prototyped an on-demand data delivery service with the market closing data. Automation tools to obtain this type of return are an increasingly common practice in this market.
The information was only available at the end of the day and in a fragmented way. It was necessary to compile data from 15,000 sources for different types of users. Because of this, there was a lack of synchrony between the closing of markets and the performance of analysts. The previous model also required a specific task of cross-checking data by IT teams – which generated long working hours.
MJV combined Design Thinking with Data Science to prototype the solution for the customer. The work was divided into three parts:
- Immersion, with professionals on the stock exchange in partnership with the innovation team, to understand how the data was worked
- Clustering of data to identify needs, which generated a co-creation workshop between MJV and the innovation team
- Prototyping the project, building and validating the digital product purchase journey to perform the analysis.
With the union between Design Thinking and Data Science, we structured and polished the creation of the product with a focus on the consumer. This way, it was possible to develop an innovative platform that delivered data in a segmented and customizable way. This project left a real legacy on the Stock Exchange and inaugurated new guidelines for the organization’s digital products, encouraging digital transformation.
What did you think of our Data Science cases? These were just four examples, but MJV has extensive know-how and a Data Science team dedicated exclusively to help companies like yours discover unlimited possibilities.
Want to learn more? Get in touch!Back