Duration: 3 days
Prerequisites: Basic knowledge of programming with Python
Objectives
- Understanding machine learning and its subfields
- Getting used to common machine learning algorithms
- Making the right choice about what algorithm to use depending on the case
- Acquiring expertise in analyzing algorithms results and performance metrics
Content
- Introduction
- Supervised learning, unsupervised learning and reinforcement learning
- Classification, regression, structural prediction
- Model evaluation: metrics
- Hyperparameter selection, model selection
- Initiation to Scikit-learn
- Data types and methods selection guide
- Classification: Introduction with Optical Character Recognition (OCR)
- K nearest neighbors’ algorithm (KNN)
- Decision trees
- Ensemble methods
- Support Vector Machines (SVM)
- Results visualization
- Classification: Advanced concepts with sentiment analysis
- Data preprocessing for learning algorithms
- Dimensionality reduction
- Batchwise training
- Interpretability
- Regression
- Linear regression
- Non-linear regression with kernel methods
- Outlier detection and handling
- Time series: Challenges, decomposition and predictive methods
- Time series: non-stationary regression and auto-regressive models
- Recommendation systems, case study
- Collaborative filtering per user
- Collaborative filtering per item
- Advanced concepts and algorithms
- Unsupervised learning
- Clustering: K-means, hierarchical clustering, density methods
- Dimensionality reduction: PCA, t-SNE,
- Generative models: Introduction to autoencoders and variational autoencoders
- Practical debugging guide
- Overfitting test
- Data pipelines test
- Exploration of alternative metrics
En savoir plus