Machine learning (ML) is a powerful tool that enables computers to learn from data and make predictions or decisions. But not all machine learning is the same – there are different types of learning, each suitable for specific tasks. The two most common types are supervised learning and unsupervised learning. In this article, we'll explore the differences between them, provide real-world examples, and walk through code snippets to help you understand how they work.
What is supervised learning?
Supervised learning is a type of machine learning in which an algorithm learns from labeled data. In other words, the data you provide to the model includes input features and the correct outputs (labels). The goal is for the model to learn the relationship between inputs and outputs so that it can make accurate predictions on new, unseen data.
Real world examples of supervised learning
Email Spam Detection:
- Input: The text of the email.
- Output: Label indicating whether the email is "Spam" or "Not Spam".
- The model learns to classify emails based on labeled examples.
House Price Forecast:
- Input: Characteristics of the home (e.g. square footage, number of bedrooms, location).
- Output: Price of the house.
- The model learns to predict prices based on historical data.
Medical Diagnosis:
- Input: Patient data (e.g., symptoms, lab results).
- Output: Diagnosis (e.g. "Health" or "Diabetes").
- The model learns to diagnose based on labeled medical records.
What is unsupervised learning?
Unsupervised learning is a type of machine learning in which algorithms learn from unlabeled data. Unlike supervised learning, no correct output is provided. Instead, models try to find patterns, structures, or relationships in the data on their own.
Real world examples of unsupervised learning
Customer segmentation:
- Input: Customer data (e.g. age, purchase history, location).
- Output: Groups of similar customers (e.g., "high-frequency buyers", "budget shoppers").
- The model identifies clusters of customers with similar behavior.
Anomaly detection:
- Input: network traffic data.
- Output: Identify unusual patterns that may indicate a cyber attack.
- The model detects outliers or anomalies in the data.
Market Basket Analysis:
- Input: Grocery store transaction data.
- Output: Groups of products that are often purchased together (e.g., "bread and butter").
- The model identifies associations between products.
The main differences between supervised learning and unsupervised learning
**方面** | **監(jiān)督學(xué)習(xí)** | **無(wú)監(jiān)督學(xué)習(xí)** |
---|---|---|
**數(shù)據(jù)** | 標(biāo)記的(提供輸入和輸出) | 未標(biāo)記的(僅提供輸入) |
**目標(biāo)** | 預(yù)測(cè)結(jié)果或?qū)?shù)據(jù)進(jìn)行分類 | 發(fā)現(xiàn)數(shù)據(jù)中的模式或結(jié)構(gòu) |
**示例** | 分類、回歸 | 聚類、降維 |
**復(fù)雜性** | 更容易評(píng)估(已知輸出) | 更難評(píng)估(沒(méi)有基本事實(shí)) |
**用例** | 垃圾郵件檢測(cè)、價(jià)格預(yù)測(cè) | 客戶細(xì)分、異常檢測(cè) |
Code Example
Let’s dig into some code and see how supervised and unsupervised learning work in practice. We will use Python and the popular Scikit-learn library.
Supervised Learning Example: Predicting House Prices
We will use a simple linear regression model to predict the price of a home based on characteristics such as square footage.
# 導(dǎo)入庫(kù) import numpy as np import pandas as pd from sklearn.model_selection import train_test_split from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error # 創(chuàng)建樣本數(shù)據(jù)集 data = { 'SquareFootage': [1400, 1600, 1700, 1875, 1100, 1550, 2350, 2450, 1425, 1700], 'Price': [245000, 312000, 279000, 308000, 199000, 219000, 405000, 324000, 319000, 255000] } df = pd.DataFrame(data) # 特征 (X) 和標(biāo)簽 (y) X = df[['SquareFootage']] y = df['Price'] # 將數(shù)據(jù)分成訓(xùn)練集和測(cè)試集 X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42) # 訓(xùn)練線性回歸模型 model = LinearRegression() model.fit(X_train, y_train) # 做出預(yù)測(cè) y_pred = model.predict(X_test) # 評(píng)估模型 mse = mean_squared_error(y_test, y_pred) print(f"均方誤差:{mse:.2f}")
Unsupervised Learning Example: Customer Segmentation
We will use K-means clustering algorithm to group customers based on their age and spending habits.
# 導(dǎo)入庫(kù) import numpy as np import pandas as pd from sklearn.cluster import KMeans import matplotlib.pyplot as plt # 創(chuàng)建樣本數(shù)據(jù)集 data = { 'Age': [25, 34, 22, 45, 32, 38, 41, 29, 35, 27], 'SpendingScore': [30, 85, 20, 90, 50, 75, 80, 40, 60, 55] } df = pd.DataFrame(data) # 特征 (X) X = df[['Age', 'SpendingScore']] # 訓(xùn)練 K 均值聚類模型 kmeans = KMeans(n_clusters=3, random_state=42) df['Cluster'] = kmeans.fit_predict(X) # 可視化集群 plt.scatter(df['Age'], df['SpendingScore'], c=df['Cluster'], cmap='viridis') plt.xlabel('年齡') plt.ylabel('消費(fèi)評(píng)分') plt.title('客戶細(xì)分') plt.show()
When to use supervised learning vs. unsupervised learning
When to use supervised learning:
- You have labeled data.
- You want to predict outcomes or classify data.
- Examples: Predicting sales, classifying images, detecting fraud.
When to use unsupervised learning:
- You have unlabeled data.
- You want to discover hidden patterns or structures.
- Examples: Group customers, reduce data dimensions, and find anomalies.
Conclusion
Supervised learning and unsupervised learning are two basic methods in machine learning, each with its own advantages and use cases. Supervised learning is great for making predictions when you have labeled data, while unsupervised learning is great when you want to explore and discover patterns in unlabeled data.
By understanding the differences and practicing with real-world examples, such as the ones in this article, you will master these basic machine learning techniques. If you have any questions or want to share your own experiences, please feel free to leave a comment below.
The above is the detailed content of Supervised vs. Unsupervised Learning. For more information, please follow other related articles on the PHP Chinese website!

Hot AI Tools

Undress AI Tool
Undress images for free

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Clothoff.io
AI clothes remover

Video Face Swap
Swap faces in any video effortlessly with our completely free AI face swap tool!

Hot Article

Hot Tools

Notepad++7.3.1
Easy-to-use and free code editor

SublimeText3 Chinese version
Chinese version, very easy to use

Zend Studio 13.0.1
Powerful PHP integrated development environment

Dreamweaver CS6
Visual web development tools

SublimeText3 Mac version
God-level code editing software (SublimeText3)

Hot Topics

Python's unittest and pytest are two widely used testing frameworks that simplify the writing, organizing and running of automated tests. 1. Both support automatic discovery of test cases and provide a clear test structure: unittest defines tests by inheriting the TestCase class and starting with test\_; pytest is more concise, just need a function starting with test\_. 2. They all have built-in assertion support: unittest provides assertEqual, assertTrue and other methods, while pytest uses an enhanced assert statement to automatically display the failure details. 3. All have mechanisms for handling test preparation and cleaning: un

Python's default parameters are only initialized once when defined. If mutable objects (such as lists or dictionaries) are used as default parameters, unexpected behavior may be caused. For example, when using an empty list as the default parameter, multiple calls to the function will reuse the same list instead of generating a new list each time. Problems caused by this behavior include: 1. Unexpected sharing of data between function calls; 2. The results of subsequent calls are affected by previous calls, increasing the difficulty of debugging; 3. It causes logical errors and is difficult to detect; 4. It is easy to confuse both novice and experienced developers. To avoid problems, the best practice is to set the default value to None and create a new object inside the function, such as using my_list=None instead of my_list=[] and initially in the function

Python works well with other languages ??and systems in microservice architecture, the key is how each service runs independently and communicates effectively. 1. Using standard APIs and communication protocols (such as HTTP, REST, gRPC), Python builds APIs through frameworks such as Flask and FastAPI, and uses requests or httpx to call other language services; 2. Using message brokers (such as Kafka, RabbitMQ, Redis) to realize asynchronous communication, Python services can publish messages for other language consumers to process, improving system decoupling, scalability and fault tolerance; 3. Expand or embed other language runtimes (such as Jython) through C/C to achieve implementation

Python's list, dictionary and collection derivation improves code readability and writing efficiency through concise syntax. They are suitable for simplifying iteration and conversion operations, such as replacing multi-line loops with single-line code to implement element transformation or filtering. 1. List comprehensions such as [x2forxinrange(10)] can directly generate square sequences; 2. Dictionary comprehensions such as {x:x2forxinrange(5)} clearly express key-value mapping; 3. Conditional filtering such as [xforxinnumbersifx%2==0] makes the filtering logic more intuitive; 4. Complex conditions can also be embedded, such as combining multi-condition filtering or ternary expressions; but excessive nesting or side-effect operations should be avoided to avoid reducing maintainability. The rational use of derivation can reduce

PythonisidealfordataanalysisduetoNumPyandPandas.1)NumPyexcelsatnumericalcomputationswithfast,multi-dimensionalarraysandvectorizedoperationslikenp.sqrt().2)PandashandlesstructureddatawithSeriesandDataFrames,supportingtaskslikeloading,cleaning,filterin

To implement a custom iterator, you need to define the __iter__ and __next__ methods in the class. ① The __iter__ method returns the iterator object itself, usually self, to be compatible with iterative environments such as for loops; ② The __next__ method controls the value of each iteration, returns the next element in the sequence, and when there are no more items, StopIteration exception should be thrown; ③ The status must be tracked correctly and the termination conditions must be set to avoid infinite loops; ④ Complex logic such as file line filtering, and pay attention to resource cleaning and memory management; ⑤ For simple logic, you can consider using the generator function yield instead, but you need to choose a suitable method based on the specific scenario.

Dynamic programming (DP) optimizes the solution process by breaking down complex problems into simpler subproblems and storing their results to avoid repeated calculations. There are two main methods: 1. Top-down (memorization): recursively decompose the problem and use cache to store intermediate results; 2. Bottom-up (table): Iteratively build solutions from the basic situation. Suitable for scenarios where maximum/minimum values, optimal solutions or overlapping subproblems are required, such as Fibonacci sequences, backpacking problems, etc. In Python, it can be implemented through decorators or arrays, and attention should be paid to identifying recursive relationships, defining the benchmark situation, and optimizing the complexity of space.

Python's regular expressions provide powerful text processing capabilities through the re module, which can be used to match, extract and replace strings. 1. Use re.search() to find whether there is a specified pattern in the string; 2. re.match() only matches from the beginning of the string, re.fullmatch() needs to match the entire string exactly; 3. re.findall() returns a list of all non-overlapping matches; 4. Special symbols such as \d represents a number, \w represents a word character, \s represents a blank character, *, , ? represents a repeat of 0 or multiple times, 1 or multiple times, 0 or 1 time, respectively; 5. Use brackets to create a capture group to extract information, such as separating username and domain name from email; 6
