Data Science & Machine Learning Data Manipulation with NumPy and Pandas Syllabus

Data Manipulation with NumPy and Pandas syllabus — full chapter and topic list

Expand any chapter to see its topics and sub-topics. This is the whole examinable outline for Data Manipulation with NumPy and Pandas in Data Science & Machine Learning, not a summary of it.

1. 1

   Numerical Computing with NumPy

   4 topics

   • ndarray Fundamentals
     • Array creation and dtypes
     • Shape, reshape and dimensions
   • Indexing and Slicing
     • Boolean and fancy indexing
   • Vectorized Operations
     • Broadcasting rules
     • Universal functions
   • Aggregation and Linear Algebra
     • sum, mean, axis operations
     • Matrix operations with np.linalg
2. 2

   Pandas Data Structures

   3 topics

   • Series and DataFrame
     • Creating from lists, dicts and arrays
     • Index objects
   • Data Selection
     • loc and iloc
     • Boolean filtering
   • Reading and Writing Data
     • CSV, Excel and SQL sources
     • JSON and parquet
3. 3

   Data Wrangling

   5 topics

   • Handling Missing Data
     • isnull, dropna and fillna
     • Interpolation
   • Grouping and Aggregation
     • groupby mechanics
     • agg, transform and apply
   • Merging and Joining
     • merge, join and concat
     • Join types
   • Reshaping Data
     • pivot and pivot_table
     • melt, stack and unstack
   • Time Series Handling
     • Datetime indexing
     • Resampling and rolling windows
4. 4

   Data Cleaning and Feature Preparation

   3 topics

   • Duplicate and Outlier Treatment
     • drop_duplicates
     • Outlier detection with IQR
   • Type Conversion and Mapping
     • astype and map
     • apply and applymap
   • String and Categorical Cleaning
     • str accessor methods
     • Category dtype

Data Manipulation with NumPy and Pandas flashcards for Data Science & Machine Learning

18 of 51 cards from the Data Manipulation with NumPy and Pandas deck — real questions with worked answers.

1. What is a NumPy ndarray?

   An N-dimensional, fixed-size array of homogeneous (same dtype) elements stored in contiguous memory, enabling fast vectorized operations.

2. What do the ndarray attributes ndim, shape, size, and dtype return?

   ndim = number of dimensions; shape = tuple of sizes per dimension; size = total number of elements; dtype = data type of elements.

3. How do np.zeros, np.ones, and np.full differ?

   np.zeros fills with 0, np.ones fills with 1, and np.full fills with a specified constant value, each for a given shape.

4. What is the difference between np.arange and np.linspace?

   np.arange(start, stop, step) generates values by step (stop exclusive); np.linspace(start, stop, num) generates num evenly spaced values (stop inclusive by default).

5. Why is a NumPy array more memory- and speed-efficient than a Python list for numeric work?

   It stores homogeneous data in contiguous memory and uses compiled, vectorized C operations, avoiding per-element Python object overhead and loop interpretation.

6. What does array broadcasting do in NumPy?

   It lets arrays of different but compatible shapes be combined in arithmetic by virtually stretching dimensions of size 1 to match, without copying data.

7. State the broadcasting rule for two dimensions to be compatible.

   Comparing shapes from the trailing dimension, two dimensions are compatible if they are equal or one of them is 1.

8. In NumPy, does basic slicing (e.g. a[1:4]) return a copy or a view?

   A view — it shares memory with the original array, so modifying the slice modifies the original.

9. What is fancy (integer-array) indexing in NumPy and does it return a view or copy?

   Indexing with an array/list of integer indices (e.g. a[[0,2,4]]); it always returns a copy, not a view.

10. How does boolean mask indexing work in NumPy, e.g. a[a > 5]?

    A boolean array of the same shape selects elements where the mask is True, returning a 1-D copy of those elements.

11. What is a vectorized operation in NumPy?

    Applying an operation to entire arrays at once via compiled loops instead of explicit Python for-loops, giving concise and fast element-wise computation.

12. What is a NumPy universal function (ufunc)? Give two examples.

    A function that operates element-wise on arrays with broadcasting support, e.g. np.add and np.sqrt (also np.exp, np.sin).

13. What does the axis parameter control in NumPy aggregations like sum(axis=0) on a 2-D array?

    axis=0 aggregates down the rows (collapsing rows, result per column); axis=1 aggregates across columns (result per row).

14. Which NumPy function computes the matrix (dot) product, and what operator is equivalent?

    np.dot(A, B) computes the matrix product; the @ operator is equivalent for 2-D arrays.

15. What does np.linalg.inv compute and when does it fail?

    It computes the inverse of a square matrix; it fails (raises LinAlgError) if the matrix is singular (non-invertible / determinant zero).

16. Name the NumPy functions for determinant and for eigenvalues/eigenvectors.

    np.linalg.det for the determinant; np.linalg.eig for eigenvalues and eigenvectors.

17. What is a pandas Series?

    A one-dimensional labeled array holding data of a single dtype, with an associated index for label-based access.

18. What is a pandas DataFrame?

    A two-dimensional labeled, size-mutable tabular structure with rows (index) and columns, where each column is a Series that can have its own dtype.

See more Data Manipulation with NumPy and Pandas flashcards →

Planning Data Manipulation with NumPy and Pandas for Data Science & Machine Learning

Data Manipulation with NumPy and Pandas (Data Science & Machine Learning) FAQ