Applied Machine Learning: from data to models that work in production
Learn to choose, train, validate and deploy machine learning models on real cases, with Prof. Daniel Medina and CUANTICO's AI team.
- Language
- Taught in Spanish
- Format
- 100% online, at your own pace
- Access
- Free
- Duration
- 6 h 11 min
- Content
- 11 sections and 63 lessons, with a final capstone project
- Level
- Intermediate: requires experience in data analysis
- Instructor
- Daniel Medina, with the support of CUANTICO's AI team
What it covers
Most courses teach algorithms; this one teaches you to decide. In 6 hours and 63 lessons you go from framing the problem to a defensible model monitored in production: prediction, segmentation, anomaly detection, recommendation and time series, always weighing performance, interpretability, cost and risk.
Who it is for
- Experienced data analysts who want to move into applied machine learning.
- BI and analytics teams in companies and public agencies that need to justify modeling decisions.
- Technical professionals who already use Python or SQL and want to take models to production.
What you will learn
By the end you will be able to:
- Choose the right technique for the goal, the data and the constraints.
- Build and validate supervised and unsupervised models with reproducible processes and the right metrics.
- Compare models by performance, interpretability, cost, robustness and overfitting risk.
- Solve prediction, segmentation, anomaly, recommendation and time series cases.
- Design a defensible solution, from framing the problem to monitoring in production.
Skills
- Model selection
- Supervised models
- Unsupervised models
- Validation and metrics
- Time series
- Explainability
- MLOps
Syllabus
11 sections and 63 lessons.
| Section | Lessons | What you take away |
|---|---|---|
| 1. Map of machine learning | 1 | Overview and course roadmap |
| 2. Data for modeling | 3 | Prepare data ready for training |
| 3. Supervised models | 8 | Regression and classification in practice |
| 4. Validation and performance | 7 | The right metrics and overfitting control |
| 5. Unsupervised learning | 8 | Clustering, dimensionality reduction and patterns |
| 6. Priority use cases | 6 | Prediction, segmentation and anomalies |
| 7. Time series | 7 | Forecasting with data over time |
| 8. Applied deep learning | 8 | When to use neural networks (and when not to) |
| 9. Recommendation and personalization | 6 | Applied recommender systems |
| 10. Trust and production | 8 | Explainability, MLOps and monitoring |
| 11. Capstone project | 1 | A complete, defensible solution |
Instructor
Daniel Medina is co-founder, CEO and Chief Scientific Officer of CUANTICO and directs CuantaIA, a research group recognized by MinCiencias, Colombia's Ministry of Science, Technology and Innovation, with the support of CUANTICO's AI team, with researchers in Bogotá, Cali, Cartagena, Manizales, the United States and Spain.
Start today, for free
The course is free, online and at your own pace.