Converting between formats in SciPy - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of converting dense to CSR format

What is the output of this code that converts a dense matrix to CSR format and prints its data array?

SciPy

import numpy as np
from scipy.sparse import csr_matrix

dense = np.array([[0, 0, 1], [2, 0, 0], [0, 3, 0]])
sparse_csr = csr_matrix(dense)
print(sparse_csr.data)

A [0 0 1 2 0 0 0 3 0]

B [1 2 3]

C [1 2 3 0]

D [1 2]

💡 Hint

CSR format stores only non-zero values in the data array.

The csr_matrix.data attribute contains only the non-zero elements of the matrix in row-major order. Here, the non-zero elements are 1, 2, and 3.

❓ data_output

intermediate

Shape after converting COO to CSC format

Given a sparse matrix in COO format with shape (4, 5), what is the shape of the matrix after converting it to CSC format?

SciPy

from scipy.sparse import coo_matrix

row  = [0, 3, 1, 0]
col  = [0, 3, 1, 2]
data = [4, 5, 7, 9]
coo = coo_matrix((data, (row, col)), shape=(4, 5))
csc = coo.tocsc()
print(csc.shape)

A (4, 4)

B (5, 4)

C (4, 5)

D (5, 5)

💡 Hint

Converting between sparse formats does not change the matrix shape.

The shape of the matrix remains the same when converting between sparse formats like COO to CSC.

🔧 Debug

advanced

Identify the error when converting dense to LIL format

What error does this code raise when converting a list of lists to a LIL sparse matrix?

SciPy

from scipy.sparse import lil_matrix

dense_list = [[1, 0], [0, 2]]
sparse_lil = lil_matrix(dense_list)
print(sparse_lil)

A TypeError: expected sequence of sequences, got list of lists

B ValueError: object too deep for desired array

C AttributeError: 'list' object has no attribute 'tocsc'

D No error, prints the LIL matrix correctly

💡 Hint

Check if scipy.sparse.lil_matrix accepts list of lists as input.

scipy.sparse.lil_matrix can accept a list of lists as input and convert it to a sparse matrix without error.

🚀 Application

advanced

Best sparse format for fast row slicing

Which sparse matrix format is best suited for fast row slicing operations on large sparse matrices?

A CSR (Compressed Sparse Row)

B CSC (Compressed Sparse Column)

C COO (Coordinate)

D DOK (Dictionary of Keys)

💡 Hint

Think about how data is stored internally for efficient row access.

CSR format stores data by rows, making row slicing very efficient compared to other formats.

🧠 Conceptual

expert

Memory difference between COO and CSR formats

Which statement correctly describes the memory usage difference between COO and CSR sparse matrix formats for the same data?

A COO uses more memory because it stores row and column indices separately for each non-zero element.

B CSR uses more memory because it stores an additional index pointer array for rows.

C Both COO and CSR use the same amount of memory for the same data.

D COO uses less memory because it stores data in a compressed form unlike CSR.

💡 Hint

Consider how indices are stored in COO vs CSR formats.

COO format stores row and column indices for each non-zero element, which can use more memory than CSR that compresses row indices into a pointer array.