Scientific computing

Fatma Kahveci
Coding
Page content

SciPy

  • A collection of packages addressing many standard problem domains in scientific computing.

    • scipy.integrate
      • Numerical integration routines and differential equation solvers
    • scipy.linalg
      • Linear algebra routines and matrix decompositions extending beyond those provided in numpy.linalg
    • scipy.optimize
      • Function optimizers (minimizers) and root-finding algorithms
    • scipy.signal
      • Signal processing tools
    • scipy.sparse
      • Sparse matrices and sparse linear system solvers
    • scipy.special
      • Wrapper around SPECFUN, a Fortran library implementing many common mathematical functions, such as the gamma function.
    • scipy.stats
      • Standard continuous and discrete probability distributions (density functions, samplers, continuous distribution functions), various statistical tests, and more descriptive statistics.

scikit-learn

  • The premier general-purpose machine learning toolkit for Pythonistas.

Submodules

1. Classification

  • SVM
  • Nearest neighbours
  • Random forest
  • Logistic regression

2. Regression

  • Lasso
  • Ridge regression

3. Clustering

  • k-means
  • spectral clustering

4. Dimensionality reductions

  • PCA
  • feature selection
  • matrix factorization

5. Model selection

  • Grid search
  • cross-validation
  • metrics

6. Preprocessing

  • feature extraction
  • normalization

Series

  • A Series is a sequence of data values.
  • If a DataFrame is a table, a Series is a list.
pd.Series([30,35,40], index=['2015 Sales', '2016 Sales', '2017 Sales'], name = 'Product A')