Machine Learning and Decision Tree
Machine Learning:
Machine learning is an AI technique that allows computers to learn and develop without being explicitly programmed. The study of how to build computer programs that can access data and learn independently is referred to as machine learning.
Learning starts with observations or data, such as examples, direct experience, or teaching, to seek patterns in data and make better judgments in the future based on the examples we offer. The main objective is for computers to learn independently, without the need for human involvement, and change their behavior accordingly.
Some Machine Learning Methods:
The words "supervised" and "unsupervised" machine learning techniques are sometimes used interchangeably.
Machine learning algorithms that are supervised can use labeled examples to apply what they've learned in the past to fresh data and anticipate future events. The learning method generates an inferred function to make predictions about output values based on examining a known training dataset. Following adequate training, the system may offer goals for any new input. The learning algorithm may also compare its output to the proper, intended output and detect mistakes, allowing the model to be adjusted as needed.
When the material used to train is neither categorized nor labeled, unsupervised machine learning techniques are employed. Unsupervised learning investigates how unlabelled data may be used to infer a function that describes a hidden structure. The system does not choose the appropriate output, but it does explore the data and can infer hidden structures from unlabeled data using datasets.
Because they utilize both labeled and unlabelled data for training – usually a small quantity of labeled data and a big amount of unlabelled data – semi-supervised machine learning algorithms lie halfway between supervised and unsupervised learning. This approach can increase learning accuracy significantly in systems that employ it. Semi-supervised learning is often used when the collected, labeled data necessitates competent and appropriate resources to train/learn from it. Obtaining unlabelled data, on the other hand, usually does not need extra resources.
Reinforcement learning algorithms are a type of machine learning algorithm that interacts with its environment by creating actions and detecting faults or rewards. The essential aspects of reinforcement learning are trial and error search and delayed reward. This approach enables machines and software agents to automatically identify the best behavior for a given situation to enhance performance. For the agent to learn which action is better, simple reward feedback is necessary; this is referred to as the reinforcement signal.
Decision Tree:
The most powerful and widely used classification and prediction technique is the decision tree. A decision tree is a flowchart-like tree structure in which each internal node represents an attribute test, each branch indicates the test's result, and each leaf node (terminal node) carries a class label.
Construction of the Decision Tree:
A tree can be "learned" by dividing the source into subgroups based on an attribute value test. Recursive partitioning is the process of repeating this procedure for each derived subset. The recursion is complete when all subsets at a node have the same value of the target variable or splitting no longer adds value to the predictions.
Because the design of a decision tree classifier does not need domain expertise or parameter setup, it is suitable for exploratory knowledge discovery. High-dimensional data is no problem for decision trees. In general, the accuracy of the decision tree classifier is good. The use of decision trees to gain categorization information is a common inductive method.
The Decision Tree Approach's Strengths and Weaknesses:
The following are some of the advantages of using decision tree methods: Decision trees can produce easy to grasp rules.
Without requiring a lot of computing, decision trees may do categorization.
Both continuous and categorical variables may be dealt with using decision trees.
Decision trees show which fields are crucial for categorization or prediction.
Decision tree techniques have the following flaws:
When estimating the value of a continuous characteristic, decision trees are less suitable.
In classification problems with a large number of classes and a limited number of training samples, decision trees are prone to mistakes.
The training of a decision tree can take a long time. Growing a decision tree takes a lot of time and resources. Before the optimal split can be determined at each node, each potential splitting field must be sorted. Combinations of fields are employed in various algorithms, and finding the best-combined weights is a challenge. Because numerous candidate sub-trees must be generated and compared, pruning methods can be costly.
Three decision trees are discussed in this section, and their names are as follows:
ID3 (Iterative Dichotomize 3)
C4.5 Algorithm
K-Nearest Neighbour Algorithm
.png)
.png "GCE Advanced Level Results 2021 PDF Download: A Level")
