Chapter 1. Preliminaries

Essential Python Libraries

• NumPy

• pandas

• matplotlib

• SciPy

NumPy

fundamental package for scientific computing

contains a powerful N-dimensional array object, also as the primary container for data to be passed between algorithms regards to data analysis

• reading/writing array-based data sets to disk

• element-wise computations with array
• mathematical operations between arrays
• linear algebra, Fourier transform, random number generation
• tools for integrating lower-level language can operate on the data stored in NumPy array without coping/modifying any data

pandas by Wes Mckinney

rich data structures (DataFrame, Index and Series etc.) and functions

primary data structure/object in pandas is DataFrame, which is 2-dimensional tabular and column oriented

well-suited for financial data: time series, which is exactly Wes Mckinney create pandas for

• NumPy: array computing features

• spreadsheet: flexible data manipulation
• rational database, such as SQL
• sophisticated indexing functionality: easy to reshape, slice, dice, aggregations, select subsets

matplotlib by John D. Hunter

producing plots and other 2D data visualizations

SciPy

collection of packages for scientific computation

• scipy.integrate: integrate/differential

• scipy.linalg: linear algebra, matrix decompositions beyond numpy.linalg
• scipy.optimize: function optimazers/minimizers, root finding
• scipy.signal: signal processing
• scipy.sparse: sparse matrices, sparse linear system
• scipy.special: wrapper around SPECFUN, a Fortran library implementing many common mathematical functions, such as the gamma function
• scipy.stats: probability distributions, statistical tests
• scipy.weave: tools for using inline C++ to accelerate array computations