I719 Fundamentals of Python/lecture6

From ICO wiki

Lists and Arrays

Part 1: List manipulation

Please see, you should know how to do these things!
https://wiki.itcollege.ee/index.php/I719_Fundamentals_of_Python/lists

Interlude: Packages and Requirements

python application often include a requirements.txt that lists pip dependencies

python packages are formatted as

package_name/
    package_name/
        __init__.py
    setup.py

and setup.py include information on depedencies
see: https://github.com/pypa/sampleproject/blob/master/setup.py

NOTE

Numpy and Pandas will not be tested on and are not required to know to continue in the class. These packages will only be used in this lecture

Numpy

  • it is faster
  • it allows multidimensional arrays
  • meant for math

Basics

import numpy as np

Matrix multiplication

a 2x2 matrix by a 2x1 column vector

[[1, 2]  . [[1],  = [[1 * 1 + 2 * 2], = [[5],
 [3, 4]]    [2]]     [3 * 1 + 4 * 2]]    [11]]

Results in a 2x1 column vector

in numpy:

In [2]: A = np.array([[1,2],[3,4]])

In [3]: V = np.array([[1],[2]])

In [4]: A.dot(V)
Out[4]:
array([[ 5],
       [11]])

Solve system of linear equations

x - y = 3
7x - y = -3

becomes in matrices
(see how: https://www.khanacademy.org/math/precalculus/precalc-matrices/solving-equations-with-inverse-matrices/v/matrix-equations-systems)

[[1, -1]  . [[x],  = [[3],
 [7, -1]]    [y]]     [-3]]

This can go back to the original equations with variables:

[[1, -1]  . [[x],  = [[1x + -1y], = [[3],
 [7, -1]]    [y]]     [7x + -1y]]    [-3]]

Solving in numpy

In [68]: a = np.array([[1, -1],[7, -1]])

In [69]: a
Out[69]:
array([[ 1, -1],
       [ 7, -1]])

In [70]: b = np.array([[3],[-3]])

In [71]: b
Out[71]:
array([[ 3],
       [-3]])

In [72]: a_inverse = np.linalg.inv(a)

In [73]: a_inverse
Out[73]:
array([[-0.16666667,  0.16666667],
       [-1.16666667,  0.16666667]])

In [74]: a_inverse.dot(b)
Out[74]:
array([[-1.],
       [-4.]])

Task 3

solve for x,y, and z

x + y + z = 6
2y + 5z = -4
2x + 5y - z = 27

Pandas

https://blockchain.info/charts/market-price?timespan=1year

download the csv

Open the the csv with pandas

import pandas as pd
btc_price = pd.read_csv('~/Downloads/market-price.csv', names=['date', 'usd'], parse_dates=[0])

TASKS

show only days where price is above 1000USD

In [4]: btc_price[btc_price.usd > 1000]

How many days was the price above 1000USD?

len(btc_price[btc_price.usd > 1000])

What was the price 6 months ago?

In [3]: import datetime as dt

In [4]: dt.datetime.utcnow() - dt.timedelta(days=365/2)
Out[4]: datetime.datetime(2016, 9, 8, 5, 12, 52, 204106)

In [5]: datetime_to_check = dt.datetime.utcnow() - dt.timedelta(days=365/2)

In [6]: date_to_check = datetime_to_check.date()

In [7]: date_to_check
Out[7]: datetime.date(2016, 9, 8)
...
In [9]: btc_price[btc_price.date == date_to_check]
Out[9]:
      datetime         usd
184 2016-09-08  627.777875

What was the average price last year?

[10]: btc_price.usd.mean()

What was the average price in august?

In [27]: august_end = dt.date(month=9, year=2016, day=1)

In [28]: august_start = dt.date(month=8, year=2016, day=1)

In [29]: btc_price[(august_start <= btc_price.date) & (btc_price.date < august_end)]
Out[29]:
          date         usd
146 2016-08-01  606.322343
...
176 2016-08-31  574.827937

In [30]: btc_price[(august_start <= btc_price.date) & (btc_price.date < august_end)].mean()
Out[30]:
usd    579.744405
dtype: float64

Plot the days on a line graph

import matplotlib.pyplot as plt
import pandas as pd

btc_price = pd.read_csv('~/Downloads/market-price.csv', names=['date', 'usd'], parse_dates=[0])



x = btc_price.date
y = btc_price.usd
plt.plot(x, y)

plt.xlabel('Date')
plt.ylabel('USD')
plt.title('BTC to USD')
plt.grid(True)
plt.show()

More information on Data Science in python

see https://github.com/jakevdp/PythonDataScienceHandbook/