%load_ext watermark 
%watermark -d -v -a 'Sebastian Raschka' -p numpy,pandas

import pandas as pd
import numpy as np

df = pd.DataFrame()

for col in ('a', 'b', 'c', 'd'):
    df[col] = pd.Series(range(1000), index=range(1000))

df.tail()

df.loc[:, ['a', 'c', 'd']].sum(axis=0)

# 1
%timeit -n 1000 -r 5 df.loc[:, ['a', 'c', 'd']].sum(axis=0)

# 2
%timeit -n 1000 -r 5 df[['a', 'c', 'd']].sum(axis=0)

# 3
%timeit -n 1000 -r 5 df[['a', 'c', 'd']].values.sum(axis=0)

[df[col].values.sum(axis=0) for col in ('a', 'c', 'd')]

# 4
%timeit -n 1000 -r 5 [df[col].values.sum(axis=0) for col in ('a', 'c', 'd')]

from numpy import einsum
[einsum('i->', df[col].values) for col in ('a', 'c', 'd')]

# 5
%timeit -n 1000 -r 5 [einsum('i->', df[col].values) for col in ('a', 'c', 'd')]

import timeit
import random
from numpy import einsum
import pandas as pd

def run_loc_sum(df):
    return df.loc[:, ['a', 'c', 'd']].sum(axis=0)

def run_einsum(df):
    return [einsum('i->', df[col].values) for col in ('a', 'c', 'd')]

orders = [10**i for i in range(4, 8)]
loc_res = []
einsum_res = []

for n in orders:

    df = pd.DataFrame()
    for col in ('a', 'b', 'c', 'd'):
        df[col] = pd.Series(range(n), index=range(n))
    
    print('n=%s (%s of %s)' %(n, orders.index(n)+1, len(orders)))

    loc_res.append(min(timeit.Timer('run_loc_sum(df)' , 
            'from __main__ import run_loc_sum, df').repeat(repeat=5, number=1)))

    einsum_res.append(min(timeit.Timer('run_einsum(df)' , 
            'from __main__ import run_einsum, df').repeat(repeat=5, number=1)))

print('finished')

%matplotlib inline

from matplotlib import pyplot as plt

def plot_1():
    
    fig = plt.figure(figsize=(12,6))
    
    plt.plot(orders, loc_res, 
             label="df.loc[:, ['a', 'c', 'd']].sum(axis=0)", 
             lw=2, alpha=0.6)
    plt.plot(orders,einsum_res, 
             label="[einsum('i->', df[col].values) for col in ('a', 'c', 'd')]", 
             lw=2, alpha=0.6)

    plt.title('Pandas Column Sums', fontsize=20)
    plt.xlim([min(orders), max(orders)])
    plt.grid()

    #plt.xscale('log')
    plt.ticklabel_format(style='plain', axis='x')
    plt.legend(loc='upper left', fontsize=14)
    plt.xlabel('Number of rows', fontsize=16)
    plt.ylabel('time in seconds', fontsize=16)
    
    plt.tight_layout()
    plt.show()
    
plot_1()

import timeit
import random
from numpy import einsum
import pandas as pd

def run_loc_sum(df, n):
    return df.loc[:, 0:n-1].sum(axis=0)

def run_einsum(df, n):
    return [einsum('i->', df[col].values) for col in range(0,n-1)]

orders = [10**i for i in range(2, 5)]
loc_res = []
einsum_res = []

for n in orders:

    df = pd.DataFrame()
    for col in range(n):
        df[col] = pd.Series(range(1000), index=range(1000))
    
    print('n=%s (%s of %s)' %(n, orders.index(n)+1, len(orders)))

    loc_res.append(min(timeit.Timer('run_loc_sum(df, n)' , 
            'from __main__ import run_loc_sum, df, n').repeat(repeat=5, number=1)))

    einsum_res.append(min(timeit.Timer('run_einsum(df, n)' , 
            'from __main__ import run_einsum, df, n').repeat(repeat=5, number=1)))

print('finished')

from matplotlib import pyplot as plt

def plot_2():
    
    fig = plt.figure(figsize=(12,6))
    
    plt.plot(orders, loc_res, 
             label="df.loc[:, 0:n-1].sum(axis=0)", 
             lw=2, alpha=0.6)
    plt.plot(orders,einsum_res, 
             label="[einsum('i->', df[col].values) for col in range(0,n-1)]", 
             lw=2, alpha=0.6)

    plt.title('Pandas Column Sums', fontsize=20)
    plt.xlim([min(orders), max(orders)])
    plt.grid()

    #plt.xscale('log')
    plt.ticklabel_format(style='plain', axis='x')
    plt.legend(loc='upper left', fontsize=14)
    plt.xlabel('Number of columns', fontsize=16)
    plt.ylabel('time in seconds', fontsize=16)
    
    plt.tight_layout()
    plt.show()
    
plot_2()