NotesByLex

Week 1 - Welcome

Lex Toumbourou · · revised

Course case studies

  • Case Study 1: Predicting house prices
    • Data:
      • prices of other houses.
      • other features of houses (number of bedrooms etc).
    • ML method: Regression
  • Case Study 2: Sentiment analysis
    • Data:
      • look at other reviews.
      • "decision boundary" -on
    • ML method: Classification
  • Case Study 3: Document retrieval
    • Data:
      • collection of articles.
    • ML Method: Clustering
  • Case Study 4: Product recommendation
    • Data:
      • past purchases
    • ML Method: Matrix Factorization
  • Case Study 5: Visual product recommender
    • Data:
      • input images (shoes)
    • ML Method: Deep Learning

2. Regression

  • Linear regression
  • Regularization:
  • Algorithms
    • Gradient descent
    • Coordinate descent
  • Concepts
    • Loss functions
    • Bias-variance tradeoff
    • Cross-validation
    • Sparsity
    • Overfitting
    • Model selection

3. Classification

  • Models
    • Linear classifiers
    • Kernels
    • Decision trees
  • Algorithms
    • Stochastic gradient descent
    • Boosting
  • Concepts
    • Decision boundaries
    • MLE
    • Ensemble methods
    • Random forest
    • CART
    • Online learning

4. Clustering & Retrieval

  • Models
    • Nearest neighbors
    • Clustering, mixtures of Gaussians
    • Latent Dirichlet allocation (LDA)
  • Algorithms
    • KD-trees, locality-sensitive hash (LSH)
    • K-means
    • Expectation-maximization (EM)
  • Concepts
    • Distance metrics
    • Approximation algorithms
    • Hashing
    • Sampling algorithms
    • Scaling up with map-reduce

5. Matrix factorizatoin & Dimensionality Reduction

  • Models
  • Algorithms
    • Coordinate descent
    • Eigen decomposition
    • SVD
  • Concepts
    • Matrix completion
    • Eigenvalues
    • Random projects
    • Cold-start problem
    • Diversity
    • Scaling up

6. Capstone: intelligent application using deep learning

  • Build & deploy a recommender using product images and text sentiment.

Goals of course

  • Minimum prereqs required to understand the concepts
  • Target audience
    • Software engs
    • Data enthusiast
    • Scientist
  • Math background
    • Basic calculus (lol)
      • Concept of derivatives
    • Basic linear algebra
      • Vectors
      • Matrices
      • Matrix multiplies
  • Programming experience
    • Basic Python

Notes about Graphlab

  • Loading data set (somewhat similiar to Panda's dataframe, I guess?)

      import graphlab
  sf = graphlab.SFrame('people-example.csv')

  • Basic operations

>>> # View dataset
>>> sf
+------------+-----------+---------------+-----+
| First Name | Last Name |    Country    | age |
+------------+-----------+---------------+-----+
|    Bob     |   Smith   | United States |  24 |
|   Alice    |  Williams |     Canada    |  23 |
|  Malcolm   |    Jone   |    England    |  22 |
|   Felix    |   Brown   |      USA      |  23 |
|    Alex    |   Cooper  |     Poland    |  23 |
|    Tod     |  Campbell |      USA      |  22 |
|   Derek    |    Ward   |  Switzerland  |  25 |
+------------+-----------+---------------+-----+
[7 rows x 4 columns]

>>> # Make histogram of ages (open in browser)
>>> sf['age'].show(view='Categorical')

>>> # Get mean age
>>> sf['age'].mean()

>>> # Apply transform function to data set
>>> transform_country = (
      lambda country: 'United States' if country == 'USA' else country_
>>> sf['Country'] = sf['Country'].apply(transform_country)