Python Mentorship Program Lecture Notes

Introduction and Admin

Welcome and program update: Completed month 1 (November) covering Python.
Feedback Form: Provided a link for feedback to improve the program.
New initiative: One-on-one doubt clearance via Zoom introduced, fill Google Form for booking.
Membership move: Request to cancel YouTube membership and switch to the new website subscription; YouTube auto-deducts money which cannot be refunded.

Continuation from last session on NumPy fundamentals, advancing into conceptual parts.
Hands-on session with key topics including NumPy arrays, comparisons, indexing, broadcasting, and plotting.

Speed of Execution: NumPy is almost 50 times faster than Python lists due to its use of C-type arrays.
Memory Usage: NumPy uses less memory, and memory usage can be optimized further by using different integer types like int16, int32, etc.
Convenience: NumPy provides a more convenient syntax for array operations.

Fancy Indexing: Allows extraction of array elements using lists of indices.

a = np.array([...])
b = a[[0, 2, 3]]  # extract specific rows like 0th, 2nd, and 3rd

Boolean Indexing: Extracts elements based on conditions.

mask = a > 50  # condition
result = a[mask]  # filtered array

Example combining conditions:

mask = (a > 50) & (a % 2 == 0)
result = a[mask]

Concept: Different-shaped arrays can be used together in arithmetic operations if they satisfy broadcasting rules.
Rules:
1. Make arrays have the same number of dimensions by prepending 1s to the smaller array.
2. Compare shapes: if sizes differ but one is 1, stretch it.
3. If dimensions mismatch and none is 1, broadcasting fails.

Sigmoid Function:

def sigmoid(x):
    return 1 / (1 + np.exp(-x))
result = sigmoid(np.array([1, 2, 3]))

Mean Square Error (MSE):

def mse(actual, predicted):
    return np.mean((actual - predicted) ** 2)
result = mse(np.array([...]), np.array([...]))

Filtering out NaN values:

result = a[~np.isnan(a)]  # removes NaNs

Basic Plotting: Plotting simple functions using Matplotlib.

import matplotlib.pyplot as plt
x = np.linspace(-10, 10, 100)
y = x
plt.plot(x, y)  # linear line
plt.show()

Example for quadratic:

y = x ** 2
plt.plot(x, y)
plt.show()

Session Recap: Covered NumPy vs Python list, advanced indexing, broadcasting, custom mathematical functions, handling missing values, and basic plotting with Matplotlib.
Feedback Reminder: Fill out the provided feedback form to help improve the program.