exercises.md

Exercises for Day 5

Using SciPy, Scikit-Learn and Pandas

1. Scipy

Linear Algebra

Have a look at the scipy.linalg module

a. Define a matrix A

[[1 -2 3]
 [4 5 6]
 [7 1 9]]

b. Define a vector b

[1 2 3]

c. Solve the linear system of equations A x = b

d. Check that your solution is correct by plugging it into the equation

e. Repeat steps a-d using a random 3x3 matrix B (instead of the vector b)

f. Solve the eigenvalue problem for the matrix A and print the eigenvalues and eigenvectors

g. Calculate the inverse, determinant of A

h. Calculate the norm of A with different orders

Statistics

Have a look at the scipy.stats module

a. Create a discrete random variable with poissonian distribution and plot its probability mass function (PMF), cummulative distribution function (CDF) and a histogram of 1000 random realizations of the variable

b. Create a continious random variable with normal distribution and plot its probability mass function (PMF), cummulative distribution function (CDF) and a histogram of 1000 random realizations of the variable

c. Test if two sets of (independent) random data comes from the same distribution

Hint: Have a look at the ttest_ind function

2. Pandas

For this exercise you need to have Pandas installed (you can try to install it with pip)

Tutorials are taken from https://github.com/guipsamora/pandas_exercises

a. Download the notebook food_facts.ipynb and learn how to load and display data with Pandas

b. Download the notebook army.ipynb and try yourself to use Pandas for filtering and sorting of data

c. Download the notebook alcohol.ipynb and try yourself to use Pandas for grouping of data

Bonus. Publishing of Python code

a. Take some piece of Python code (e.g. the simple_math.py module) and make it into a package.

A package should include a __init__.py inside all subdirectories, it should be able to import all the submodules from the top level, functions should have documentation (docstrings) and ideally there is a docs folder with Sphinx documentation. You should also include a small Readme.md file

b. Add a setup.py file to your package (following the instructions from the lecture notes)

Now you should be able to install your package using pip

pip install . --user

c. Upload your package to the TestPyPi repository

Follow the instructions from here