INTRODUCTION TO NUMPY doubling_array = [1, 2, 4, 8, 16, 32, 64, 128, 256, 512]
1. Your first NumPy array # Create an array of integers from one to ten
one_to_ten = [Link](1, 11)
# edited/added
# Create your scatterplot
import numpy as np
[Link](one_to_ten, doubling_array)
sudoku_list = [Link]('sudoku_game.npy')
[Link]()
# Import NumPy
import numpy as np 4. Array dimensionality
# Convert sudoku_list into an array
# edited/added
sudoku_array = [Link](sudSoku_list)
sudoku_solution = [Link]('sudoku_solution.npy')
# Print the type of sudoku_array
sudoku_list = [Link]('sudoku_game.npy')
print(type(sudoku_array))
sudoku_game = [Link](sudoku_list)
2. Creating arrays from scratch # Create the game_and_solution 3D array
game_and_solution = [Link]([sudoku_game, sudoku_solution])
# Create an array of zeros which has four columns and two rows
# Print game_and_solution
zero_array = [Link]((2, 4))
print(game_and_solution)
print(zero_array)
# Create an array of random floats which has six columns and three rows 5. The fourth dimension
random_array = [Link]((3, 6))
# edited/added
print(random_array)
new_sudoku_game = [Link]('new_sudoku_game.npy')
3. A range array new_sudoku_solution = [Link]('new_sudoku_solution.npy')
game_and_solution = [Link]('game_and_solution.npy')
doubling_array = [1, 2, 4, 8, 16, 32, 64, 128, 256, 512]
# Create a second 3D array of another game and its solution
[Link](x_values, y_values)
new_game_and_solution = [Link]([new_sudoku_game,
[Link]()
new_sudoku_solution])
# edited/addedfrom matplotlib import pyplot as plt
# Create a 4D array of both game and solution 3D arrays zero_array = [Link]((3, 2))
games_and_solutions = [Link]([game_and_solution, # Print the data type of zero_array
new_game_and_solution]) print(zero_array.dtype)
# Print the shape of your 4D array # Create an array of zeros with three rows and two columns
print(games_and_solutions.shape) zero_array = [Link]((3, 2))
6. Flattening and reshaping # Print the data type of zero_array
print(zero_array.dtype)
# edited/added
# Create a new array of int32 zeros with three rows and two columns
sudoku_game = [Link]('sudoku_game_new.npy')
zero_int_array = [Link]((3, 2), dtype=np.int32)
# Flatten sudoku_game
# Print the data type of zero_int_array
flattened_game = sudoku_game.flatten()
print(zero_int_array.dtype)
# Print the shape of flattened_game
print(flattened_game.shape) 8. Anticipating data types
# Flatten sudoku_game A string data type
flattened_game = sudoku_game.flatten()
[Link]([78.988, "NumPy", True])
# Print the shape of flattened_game [Link]([9, 1.12, True]).astype("<U5")
print(flattened_game.shape)
An integer data type
# Reshape flattened_game back to a nine by nine array
reshaped_game = flattened_game.reshape((9, 9)) [Link]([34.62, 70.13, 9]).astype(np.int64)
[Link]([45.67, True], dtype=np.int8)
# Print sudoku_game and reshaped_game
print(sudoku_game) A float data type
print(reshaped_game)
[Link]([[6, 15.7], [True, False]])
[Link]((4, 5))
7. NumPy data types
9. A smaller sudoku game
The dtype argument
# Print the data type of sudoku_game
# Create an array of zeros with three rows and two columns
print(sudoku_game.dtype) 11. Stepping into 2D
# Print the data type of sudoku_game # Create an array of the first 100 trunk diameters from tree_census
print(sudoku_game.dtype) hundred_diameters = tree_census[:100, 2]
# Change the data type of sudoku_game to int8 print(hundred_diameters)
small_sudoku_game = sudoku_game.astype(np.int8) # Create an array of trunk diameters with even row indices from 50 to 100
inclusive
# Print the data type of small_sudoku_game
every_other_diameter = tree_census[Link], 2]
print(small_sudoku_game.dtype)
print(every_other_diameter)
10. Slicing and indexing trees
12. Sorting trees
# edited/added
# Extract trunk diameters information and sort from smallest to largest
tree_census = [Link]('tree_census.npy')
sorted_trunk_diameters = [Link](tree_census[:, 2])
# Select all rows of block ID data from the second column
print(sorted_trunk_diameters)
block_ids = tree_census[:, 1]
# Print the first five block_ids 13. Filtering with masks
print(block_ids[:5])
# Create an array which contains row data on the largest tree in tree_census
# Select all rows of block ID data from the second column
largest_tree_data = tree_census[tree_census[:, 2] == 51]
block_ids = tree_census[:, 1]
print(largest_tree_data)
# Select the tenth block ID from block_ids
# Create an array which contains row data on the largest tree in tree_census
tenth_block_id = block_ids[9]
largest_tree_data = tree_census[tree_census[:, 2] == 51]
print(tenth_block_id)
print(largest_tree_data)
# Select all rows of block ID data from the second column
# Slice largest_tree_data to get only the block id
block_ids = tree_census[:, 1]
largest_tree_block_id = largest_tree_data[:, 1]
# Select five block IDs from block_ids starting with the tenth ID
print(largest_tree_block_id)
block_id_slice = block_ids[9:14]
# Create an array which contains row data on the largest tree in tree_census
print(block_id_slice)
largest_tree_data = tree_census[tree_census[:, 2] == 51]
print(largest_tree_data) Compatible along the first axis
# Slice largest_tree_data to get only the block ID (4, 2) and (6, 2)
largest_tree_block_id = largest_tree_data[:, 1] (15, 5) and (100, 5)
print(largest_tree_block_id) Compatible along the second axis
# Create an array which contains row data on all trees with
largest_tree_block_id (4, 2) and (4, 3)
trees_on_largest_tree_block = tree_census[tree_census[:, 1] ==
Not compatible
largest_tree_block_id]
print(trees_on_largest_tree_block) (5, 2) and (7, 4)
(4, 2) and (4,)
14. Fancy indexing vs. [Link]() (4, 2) and (2,)
# Create the block_313879 array containing trees on block 313879 17. Adding rows
block_313879 = tree_census[tree_census[:, 1] == 313879] # edited/added
print(block_313879) new_trees = [Link]([[1211, 227386, 20, 0], [1212, 227386, 8, 0]])
# Create an array of row_indices for trees on block 313879 # Print the shapes of tree_census and new_trees
row_indices = [Link](tree_census[:, 1] == 313879) print(tree_census.shape, new_trees.shape)
# Create an array which only contains data for trees on block 313879 # Print the shapes of tree_census and new_trees
block_313879 = tree_census[row_indices] print(tree_census.shape, new_trees.shape)
print(block_313879) # Add rows to tree_census which contain data for the new trees
updated_tree_census = [Link]((tree_census, new_trees))
15. Creating arrays from conditions
print(updated_tree_census)
# Create and print a 1D array of tree and stump diameters
trunk_stump_diameters = [Link](tree_census[:, 2] == 0, tree_census[:, 3], 18. Adding columns
tree_census[:, 2])
# Print the shapes of tree_census and trunk_stump_diameters
print(trunk_stump_diameters)
print(trunk_stump_diameters.shape, tree_census.shape)
16. Compatible or not? # Print the shapes of tree_census and trunk_stump_diameters
print(trunk_stump_diameters.shape, tree_census.shape) 20. Sales totals
# Reshape trunk_stump_diameters # edited/added
reshaped_diameters = trunk_stump_diameters.reshape((1000, 1)) monthly_sales = [Link]('monthly_sales.npy')
# Print the shapes of tree_census and trunk_stump_diameters # Create a 2D array of total monthly sales across industries
print(trunk_stump_diameters.shape, tree_census.shape) monthly_industry_sales = monthly_sales.sum(axis=1, keepdims=True)
# Reshape trunk_stump_diameters print(monthly_industry_sales)
reshaped_diameters = trunk_stump_diameters.reshape((1000, 1)) # Create a 2D array of total monthly sales across industries
# Concatenate reshaped_diameters to tree_census as the last column monthly_industry_sales = monthly_sales.sum(axis=1, keepdims=True)
concatenated_tree_census = [Link]((tree_census, print(monthly_industry_sales)
reshaped_diameters), axis=1) # Add this column as the last column in monthly_sales
print(concatenated_tree_census) monthly_sales_with_total = [Link]((monthly_sales,
monthly_industry_sales), axis=1)
19. Deleting with [Link]()
print(monthly_sales_with_total)
# Delete the stump diameter column from tree_census
tree_census_no_stumps = [Link](tree_census, 3, axis=1) 21. Plotting averages
# Save the indices of the trees on block 313879 # Create the 1D array avg_monthly_sales
private_block_indices = [Link](tree_census[:, 1] == 313879) avg_monthly_sales = monthly_sales.mean(axis=1)
# Delete the stump diameter column from tree_census print(avg_monthly_sales)
tree_census_no_stumps = [Link](tree_census, 3, axis=1) # Create the 1D array avg_monthly_sales
# Save the indices of the trees on block 313879 avg_monthly_sales = monthly_sales.mean(axis=1)
private_block_indices = [Link](tree_census[:,1] == 313879) print(avg_monthly_sales)
# Delete the rows for trees on block 313879 from tree_census_no_stumps # Plot avg_monthly_sales by month
tree_census_clean = [Link](tree_census_no_stumps, [Link]([Link](1, 13), avg_monthly_sales, label="Average sales across
private_block_indices, axis=0) industries")
# Print the shape of tree_census_clean # Plot department store sales by month
print(tree_census_clean.shape)
[Link]([Link](1, 13), monthly_sales[:, 2], label="Department store tax_collected = monthly_sales * 0.05
sales") print(tax_collected)
[Link]() # Create an array of sales revenue plus tax collected by industry and month
[Link]() total_tax_and_revenue = tax_collected + monthly_sales
22. Cumulative sales print(total_tax_and_revenue)
# Find cumulative monthly sales for each industry 24. Projecting sales
cumulative_monthly_industry_sales = monthly_sales.cumsum(axis=0) # edited/added
print(cumulative_monthly_industry_sales) monthly_industry_multipliers = [Link]('monthly_industry_multipliers.npy')
# Find cumulative monthly sales for each industry # Create an array of monthly projected sales for all industries
cumulative_monthly_industry_sales = monthly_sales.cumsum(axis=0) projected_monthly_sales = monthly_sales * monthly_industry_multipliers
print(cumulative_monthly_industry_sales) print(projected_monthly_sales)
# Plot each industry's cumulative sales by month as separate lines # Create an array of monthly projected sales for all industries
[Link]([Link](1, 13), cumulative_monthly_industry_sales[:, 0], projected_monthly_sales = monthly_sales * monthly_industry_multipliers
label="Liquor Stores")
print(projected_monthly_sales)
[Link]([Link](1, 13), cumulative_monthly_industry_sales[:, 1],
label="Restaurants") # Graph current liquor store sales and projected liquor store sales by month
[Link]([Link](1, 13), cumulative_monthly_industry_sales[:, 2], [Link]([Link](1, 13), monthly_sales[:, 0], label="Current liquor store
label="Department stores") sales")
[Link]() [Link]([Link](1, 13), projected_monthly_sales[:, 0], label="Projected
liquor store sales")
[Link]()
[Link]()
23. Tax calculations [Link]()
# Create an array of tax collected by industry and month
25. Vectorizing .upper()
tax_collected = monthly_sales * 0.05
# edited/added
print(tax_collected)
names = [Link]([["Izzy", "Monica", "Marvin"],
# Create an array of tax collected by industry and month
["Weber", "Patel", "Hernandez"]]) print(monthly_growth_2D * monthly_sales)
# Vectorize the .upper() string method
28. Broadcasting across rows
vectorized_upper = [Link]([Link])
# Apply vectorized_upper to the names array # Find the mean sales projection multiplier for each industry
uppercase_names = vectorized_upper(names) mean_multipliers = monthly_industry_multipliers.mean(axis=0)
print(uppercase_names) print(mean_multipliers)
# Find the mean sales projection multiplier for each industry
26. Broadcastable or not? mean_multipliers = monthly_industry_multipliers.mean(axis=0)
Broadcastable print(mean_multipliers)
(3, 4) and (1, 4) # Print the shapes of mean_multipliers and monthly_sales
(3, 4) and (4, ) print(mean_multipliers.shape, monthly_sales.shape)
(3, 4) and (3, 1)
# Find the mean sales projection multiplier for each industry
Not Broadcastable mean_multipliers = monthly_industry_multipliers.mean(axis=0)
print(mean_multipliers)
(3, 4) and (1, 2)
(3, 4) and (4, 1) # Print the shapes of mean_multipliers and monthly_sales
(3, 4) and (3, ) print(mean_multipliers.shape, monthly_sales.shape)
27. Broadcasting across columns # Multiply each value by the multiplier for that industry
projected_sales = monthly_sales * mean_multipliers
# edited/added
print(projected_sales)
monthly_growth_rate = [1.01, 1.03, 1.03, 1.02, 1.05, 1.03, 1.06, 1.04, 1.03,
1.04, 1.02, 1.01]
29. Loading .npy files
# Convert monthly_growth_rate into a NumPy array
monthly_growth_1D = [Link](monthly_growth_rate) # Load the mystery_image.npy file with open("mystery_image.npy", "rb") as
f:
# Reshape monthly_growth_1D
rgb_array = [Link](f)
monthly_growth_2D = monthly_growth_1D.reshape((12, 1))
# Multiply each column in monthly_sales by monthly_growth_2D
[Link](rgb_array)
[Link]() 32. Augmenting Monet
# Flip rgb_array so that it is the mirror image of the original
30. Getting help
mirrored_monet = [Link](rgb_array, axis=1)
# Display the documentation for .astype() [Link](mirrored_monet)
help([Link]) [Link]()
# Flip rgb_array so that it is upside down
31. Update and save
upside_down_monet = [Link](rgb_array, axis=(0, 1))
# edited/added [Link](upside_down_monet)
rgb_array = [Link]('rgb_array.npy') [Link]()
# Reduce every value in rgb_array by 50 percent
darker_rgb_array = rgb_array * 0.5 33. Transposing your masterpiece
# Reduce every value in rgb_array by 50 percent # Transpose rgb_array
darker_rgb_array = rgb_array * 0.5 transposed_rgb = [Link](rgb_array, axes=(1, 0, 2))
# Convert darker_rgb_array into an array of integers [Link](transposed_rgb)
darker_rgb_int_array = darker_rgb_array.astype(np.int8) [Link]()
[Link](darker_rgb_int_array)
[Link]() 34. 2D split and stack
# Reduce every value in rgb_array by 50 percent # Split monthly_sales into quarterly data
darker_rgb_array = rgb_array * 0.5 q1_sales, q2_sales, q3_sales, q4_sales = [Link](monthly_sales, 4)
# Convert darker_rgb_array into an array of integers print(q1_sales)
darker_rgb_int_array = darker_rgb_array.astype(np.int8) # Split monthly_sales into quarterly data
[Link](darker_rgb_int_array) q1_sales, q2_sales, q3_sales, q4_sales = [Link](monthly_sales, 4)
[Link]() print(q1_sales)
# Save darker_rgb_int_array to an .npy file called darker_monet.npywith # Stack the four quarterly sales arrays
open("darker_monet.npy", "wb") as f: quarterly_sales = [Link]([q1_sales, q2_sales, q3_sales, q4_sales])
[Link](f, darker_rgb_int_array)
print(quarterly_sales) # Print the shapes of blue_array and emphasized_blue_array_2D
print(blue_array.shape, emphasized_blue_array_2D.shape)
35. Splitting RGB data
# Reshape red_array and green_array
# Split rgb_array into red, green, and blue arrays red_array_2D = red_array.reshape((675, 843)) # edited/added
red_array, green_array, blue_array = [Link](rgb_array, 3, axis=2) green_array_2D = green_array.reshape((675, 843)) # edited/added
# Split rgb_array into red, green, and blue arrays # Print the shapes of blue_array and emphasized_blue_array_2D
red_array, green_array, blue_array = [Link](rgb_array, 3, axis=2) print(blue_array.shape, emphasized_blue_array_2D.shape)
# Create emphasized_blue_array # Reshape red_array and green_array
emphasized_blue_array = [Link](blue_array > blue_array.mean(), 255, red_array_2D = red_array.reshape((675, 843)) # edited/added
blue_array)
green_array_2D = green_array.reshape((675, 843)) # edited/added
# Print the shape of emphasized_blue_array
# Stack red_array_2D, green_array_2D, and emphasized_blue_array_2D
print(emphasized_blue_array.shape)
emphasized_blue_monet = [Link]([red_array_2D, green_array_2D,
# Split rgb_array into red, green, and blue arrays emphasized_blue_array_2D], axis=2)
red_array, green_array, blue_array = [Link](rgb_array, 3, axis=2) [Link](emphasized_blue_monet)
# Create emphasized_blue_array [Link]()
emphasized_blue_array = [Link](blue_array > blue_array.mean(), 255,
blue_array)
# Print the shape of emphasized_blue_array
print(emphasized_blue_array.shape)
# Remove the trailing dimension from emphasized_blue_array
emphasized_blue_array_2D = emphasized_blue_array.reshape((675, 843)) #
edited/added
36. Stacking RGB data
# Print the shapes of blue_array and emphasized_blue_array_2D
print(blue_array.shape, emphasized_blue_array_2D.shape)
