Analytics

Machine Learning models extract knowledge and predictive capacity from data. They are one of the main sources of innovation in business management.
Do you realize the value of your data?

Descriptive Models
  • Clustering
  • Association rules
  • Dimensionality reduction
Predictive Models
  • Customer churn
  • Demand forecasting
  • Price prediction
  • Text mining
  • Classification models

Examples of our work:

  • Your customers are not all the same, use clustering models to understand them better. Clustering allows you to group not only customers, but also visitors to a website or users of an application - according to common characteristics.

    If you apply a clustering model to a portfolio of customers, they will be grouped according to variables that describe their behavior: purchase frequency, purchase value, preferred type of products, most used means of purchase, sensitivity to discounts and price, etc.

    When applied in a marketing context, clustering models are an essential basis for market segmentation and the definition of high-impact differentiated marketing strategies.

    Algorithms:

    • K-Means
    • Mini-batch K-Means
    • CLARA
    • DBSCAN
  • Association rules make it possible to discover frequent sets of products. More than "cheese and ham" or "nappies and beer", these models identify products that sell better together than separately.

    These associations are not always obvious to the marketer. Discovering them is an opportunity for cross-selling and developing new products or services.

    Algorithms:

    • FP-Growth
    • Apriori
  • What are the characteristics that precede customer abandonment? How can we anticipate and prevent this abandonment?

    Churn models have predictive capacity to answer these questions. By identifying those customers at risk of churn, it is possible to act in advance, seeking to preserve customers.

    Algorithms:

    • Logistic regression
    • Random forest
    • Support Vector Machines (SVM)
  • What is the value of a property or a car based on its characteristics? Explore the market and look for opportunities to find under- or over-valued assets.

    From data with the characteristics of a property, such as its location, year of construction, area, typology, energy certificate, among others, it is possible to accurately predict its value.

    Algorithms:

    • Linear regression
    • Random forest
    • AdaBoostRegressor
  • What do your customers and followers say about your brand? When your customers complain or comment on your website, what are their main questions?

    How do you extract meaning from a vast array of documents?

    In many businesses, text remains an untapped source of value. Text stored in emails, documents or social media is also data.

Contact us

Talk to us about a business sotuation you would like to solve. If you have the data, whether tables, text or images, we have the solutions at your disposal.