Applied Machine Learning from zero

These are all the posts where I described what I have learned about data, machine learning & Python (no better way to learn a new language than coupling it with a new topic).

Focus is on applied. I will add real examples how to deploy a solution.
This is a work in progress.

Introduction: data, ML and Python

  1. Big data, data science and machine learning explained
  2. Measures of central tendency / list and tuples
  3. The median / sort a list
  4. The mode / dictionary and loops
  5. Deviation measures / power operator
  6. The coefficient of variation / exceptions
  7. Automatic documentation
  8. Rounding error
  9. Automatic tests
  10. Plotting a bar chart
  11. Plotting on the web
  12. Quartiles and summary statistics / array subsets
  13. Visualize quartiles
  14. Introduction to NumPy
  15. Introduction to Pandas
  16. Read and clean data with Python pandas
  17. Scatterplot
  18. Covariance / list comprehension
  19. Tidy Data: examples and tutorials on GitHub
  20. Example: read and format the NHL football data (a GitHub Gist)

Predict

Regression to predict continuous values

  1. Linear regression / vectors operations
  2. Gradient descent / matrix operations
  3. Linear Regression applied: Moneyball
  4. Multiple variable linear regression
  5. Inference in regression: confidence intervals and more
  6. Features selection
  7. Overfitting & bias-variance dilemma
  8. Cross-validation & regularisation
  9. Non linear regression
  10. Evaluate models
  11. Azure examples
  12. App: model & predict

Regression to predict discrete values

  1. Linear Regression with qualitative predictors
  2. Logistic regression
  3. Logistic regression with statsmodels
  4. Example: TBD (probably from healthcare)
  5. Model Pipeline

Regression to predict time series

  1. A basic example

Classify

  1. Text Processing intro / collections, set, lambda, regex, files
  2. Distributed Computing: Count Words using Apache Spark
  3. Introduction to NLTK
  4. Example: classify an article
  5. Example: sentiment analysis
  6. Example: classify a proper name
  7. K-neighbours algorithm
  8. SVC. When classification speed is more critical than training speed.
  9. App: NLP3o web app, a text analyser

Neural Networks

  1. Perceptron / generators
  2. Neural Network intro, the back-propagation
  3. A deep neural network to classify an image
  4. Example multi/classification: recognise numbers
  5. Deep Learning hyper-parameters tuning
  6. App: TBD

Connect

Cluster and similarity

  1. Unsupervised learning
  2. Example: retrieve documents

Advanced time series

  1. Bayesian models
  2. Hidden Markov Models

Interact

Recommendation systems

  1. Example: movies / books / products

Reason

  1. Software agents

Appendix: Python

  1. … / iterables, generator functions
  2. … / slice
  3. …/ functions with any number of arguments / attaching metadata
  4.  … / anonymous or inline functions, lambda
  5. …/ decorators
  6. … / Modules, classes, objects, inheritance, composition
  7. … / functional programming , map and reduce, lambda, pipelines, closure
  8. …/ concurrency, threads
  9. … / profiling and timing
Advertisements