Module 2: Creating NumPy Arrays

In this module, we'll explore the different ways to create arrays. While you can create arrays from lists, NumPy provides specialized functions to generate large datasets efficiently.

---

Lesson 3: Array Creation Methods

Basic Creation

The most direct way is using np.array(). You can also use np.asarray() which is similar but avoids copying if the input is already an array.

Initializing with Placeholders

When working with large datasets, you often need to initialize an array of a specific shape first.

• np.zeros(): Creates an array filled with 0s.
• np.ones(): Creates an array filled with 1s.
• np.empty(): Creates an array without initializing entries (faster, but contains random memory junk).
• np.full(): Creates an array filled with a specific value.

Numerical Ranges

• np.arange(start, stop, step): Like Python's range(), but returns an array.
• np.linspace(start, stop, num): Returns num evenly spaced numbers over a specified interval. Perfect for plotting!

Identity and Eye

• np.eye(n): Creates a 2D identity matrix (1s on the diagonal).
• np.identity(n): Similar to eye.

---

Lesson 4: Data Types (dtype)

Every NumPy array has a dtype. You can specify it during creation or convert it later.

Common dtypes:

• int32, int64: Integers.
• float32, float64: Floating point numbers.
• bool: True/False.
• complex128: Complex numbers.

Memory Optimization

Choosing the right dtype is crucial for large data. For example, using int8 instead of int64 can save 8x memory if your numbers are small (0-255).

---

Mini Project: Synthetic Sensor Data

Imagine you need to simulate 100 sensor readings between 20°C and 30°C.

Challenge: Use np.linspace() to generate 100 readings and then convert them to int16 using astype().

Key Takeaways

✅ Use placeholder functions like zeros() to pre-allocate memory.
✅ linspace is better for total count, arange is better for step size.
✅ Always be mindful of your dtype to optimize memory.