2.2.4. Two Subtypes of Supervised Learning

🧠 Two Subtypes of Supervised Learning

1. Regression — Output is a number E.g., Predict marks, temperature, price
2. Classification — Output is a category E.g., Spam or Not Spam, Pass or Fail

Supervised learning can be divided into two main subtypes: regression and classification. In regression, the output is a numeric value, which means we're trying to predict or estimate values based on input data. For instance, predicting a student's marks based on the hours they studied is a regression problem. In contrast, classification deals with predicting categories. This means that the output is discrete, such as classifying emails as 'spam' or 'not spam'.

Imagine you're a teacher. When you grade a student's exam, you might give them a score, such as a number from 0 to 100—this is like regression. Now, if you decide whether the student passes or fails based on their score, you're making a classification.

📌 Example 1: Regression (Predict Numbers) Let’s predict marks from hours studied.

from sklearn.linear_model import LinearRegression
import numpy as np
X = np.array([[1], [2], [3], [4], [5]]) # Hours
y = np.array([35, 45, 55, 65, 75]) # Marks
model = LinearRegression()
model.fit(X, y)
print("Prediction for 6 hours:", model.predict([[6]])[0])

🔍 Explanation: ● The model sees how marks increase with hours. ● It finds a best-fit line (like a graph) between hours and marks. ● Then it predicts marks for 6 hours using the same pattern.

In this chunk, we dive into the regression subtype with a practical example. We use Python and a library called sklearn to create a regression model that predicts student marks based on study hours. The model is trained using data that relates study hours to actual marks. After training, it can make predictions for an unseen number of study hours, like predicting the marks if a student studies for 6 hours. The algorithm identifies the relationship between the input (study hours) and output (marks) by fitting a line through the data points.

Think of regression as drawing a line through points on a graph. If you have measurements of how much someone studied and their corresponding scores, you can visualize this as dots on a graph. The straight line that best connects these dots helps you to predict scores for given study hours, similar to how you might estimate how much a plant grows based on sunlight exposure.

📌 Example 2: Classification (Predict Categories) Now let’s predict pass/fail:

from sklearn.neighbors import KNeighborsClassifier
import numpy as np
X = np.array([[1], [2], [3], [4], [5]])
y = np.array([0, 0, 1, 1, 1]) # 0 = Fail, 1 = Pass
model = KNeighborsClassifier(n_neighbors=3)
model.fit(X, y)
print("Prediction for 2.5 hours:", model.predict([[2.5]])[0])

🔍 Explanation: ● KNN checks the closest 3 students. ● If most of them passed, it predicts "pass" for 2.5 hours.

This chunk presents an example of classification using a method called K-Nearest Neighbors (KNN). Here, we want to predict whether a student passes or fails based on the hours they studied. The output is categorical—0 or 1 for fail or pass. The KNN algorithm looks at the closest training examples in the dataset to the new instance (2.5 hours in this case) and makes a prediction based on the majority category among those closest neighbors.

Imagine a group project where students who studied different amounts respond to a survey asking if they feel prepared for the exam. If most students who studied similar hours to a new student reported feeling prepared (or unprepared), you can infer whether that new student is likely to pass or fail based on their study habits. This is like KNN, where you only consider a small group (the nearest neighbors) when determining the prediction.

--