What is Machine Learning?

Machine learning (ML) is a process where we provide input to a machine, and in return, the machine generates answers or responses. However, the machine doesn't respond randomly; it relies on patterns and information provided to it in advance. It's similar to teaching a child about different fruits by showing them various examples. The child learns to distinguish between fruits based on features such as color and shape.

In the same manner, we expose the machine to different features and data. The machine comprehends each feature and input. Through a combination of mathematical and statistical approaches, and with the help of programming languages like Python, the machine learns to recognize patterns within the data. This understanding enables the machine to make predictions for future inputs. Overall, the entire process is known as machine learning, where the machine learns from data, understands patterns, and applies this knowledge to make predictions. In today's world, Artificial Intelligence (AI) is a term we hear everywhere, and it has become a significant part of our daily lives, much like our phones. AI, found in various applications, plays a crucial role in many aspects of our day-to-day activities.

AI is smart, almost like a person, and it does a lot for us. It follows our instructions, gives us feedback, and handles various tasks, such as making cars drive themselves, organizing emails, and even engaging in conversations through tools like ChatGPT.

Artificial Intelligence (AI) is like a mix of different things: Machine Learning (ML), Deep Learning (DL), and Natural Language Processing (NLP). For AI to be smart, it needs lots of data that the system learns from. This data helps AI make decisions in different situations and help us out.

In AI, ML is super important. First, we gather a bunch of data, clean it up, and try to find patterns in it. This whole process is called data engineering and data analysis. It's also a part of data science. So, in simple words, AI uses ML, and ML needs data to learn and make smart choices. Analyzing data and making predictions based on it,

this has always been done everywhere in the corporate world or in the political world. This work is being done for hundreds of years. Earlier, this work was done manually with the help of mathematics and statistics. Today, this work is done through computers. In the era when we have the best hardware and the best programming language, we can easily convert the mathematical statistics approach into programming and analyze it easily and fast, that is why today its impact is visible in every field. This process is called Machine Learning. There are different types of ML

1. Supervised learning.
2. Unsupervised Learning.
3. Semi-Supervised Learning.
4. Reinforcement Learning.
5. Deep Learning

Supervised Learning

Supervised Machine Learning is like teaching a machine by giving it examples and answers, and then letting it learn from those examples to make predictions in the future.

Imagine you're teaching a student in school. You provide them with lots of exercises along with the correct answers. The student learns from these exercises and answers, understanding the patterns and relationships between the questions and the correct responses. When they face similar questions in the exam, they can give the right answers based on what they've learned. Similarly, in Supervised Machine Learning, we give the machine lots of examples of inputs (like questions) along with the correct outputs (like answers). The machine learns from this data, figuring out the patterns and relationships between the inputs and outputs. Then, when we give it new inputs, it can predict the outputs based on its training.

For instance, if you train a machine to recognize pictures of cats by showing it lots of cat pictures and labeling them as "cat," it learns what features are common in cat pictures. Then, when you show it a new picture of a cat, it can correctly identify it as a cat.

But, if you ask it about something completely different, like a car, which it hasn't been trained on, it won't be able to give you a meaningful answer. It's like the student who, when faced with a question outside of their textbook, says, "I don't know, it's not in my book!"

So, in simple terms, supervised machine learning is all about learning from examples with known answers and then using that learning to make predictions on new data. In machine learning, data training often involves two primary patterns: linear and categorical.

Linear Data Training: Linear data training deals with establishing relationships between two variables in a linear format. These variables are typically denoted as follows:

Independent variable (X): This is the variable whose values are controlled or manipulated. It is also referred to as the predictor variable.

Dependent variable (Y): This variable's values are influenced by changes in the independent variable. It is also known as the response variable.

In linear data training, we aim to understand how changes in the independent variable (X) affect the dependent variable (Y). This relationship can exhibit two primary types of trends: a. Positive Relation: If an increase in the independent variable (X) leads to a corresponding increase in the dependent variable (Y), we say there's a positive relation between X and Y. b. Negative Relation: Conversely, if an increase in the independent variable (X) results in a decrease in the dependent variable (Y), it indicates a negative relation between X and Y. These relationships can be quantified using mathematical and statistical formulas. One common measure is the correlation coefficient, which assesses the strength and direction of the relationship between X and Y. A correlation coefficient close to +1 indicates a strong positive correlation, while a value near -1 suggests a strong negative correlation. A coefficient close to 0 indicates a weak or no linear relationship between the variables. Mathematically, the correlation coefficient (ρ) can be calculated using the covariance (Cov) between x and y, divided by the product of their standard deviations (σ_X and σ_Y): p=Cov(x,y)/σ_x*σ_y These relationships can be visually represented using a line graph. In a line graph, the values of X are plotted on the horizontal axis, and the values of Y are plotted on the vertical axis. A positive relationship is illustrated by a line that slopes upwards from left to right, while a negative relationship is depicted by a line that slopes downwards. Y is increasing according to x its positive relation Y Decreasing according to x its negative relation Categorial Data Training: In categorical machine learning (ML), we deal with data that falls into multiple categories, and our goal is to understand which features are associated with each category.

Example with Fruits:

For instance, imagine we have different types of fruits, each with its own set of features such as color, shape, size, and texture. We want the machine to learn from these features and recognize which category (or type of fruit) a particular fruit belongs to.

Training Process:

During the training process, we provide the machine with many examples of fruits along with their corresponding features. For example, we might show it an apple with features like red color, round shape, medium size, and smooth texture.

Feature Importance:

The machine then learns to recognize patterns in the data, understanding which features are important for distinguishing between different categories. For instance, it might learn that round shape and red color are common features of apples, while elongated shape and yellow color are typical of bananas.

Probability Formulas:

In categorical ML, we often use probability formulas from statistics to make predictions. These formulas help us calculate the likelihood of a particular category given the observed features. For example, we might use Bayes' theorem or logistic regression to estimate the probability that a fruit is an apple or a banana based on its color, shape, size, etc.

Prediction and Recognition:

Once the machine has been trained, we can use it to predict the category of new fruits based on their features. For example, if we provide the machine with the features of a fruit it hasn't seen before, such as green color, elongated shape, small size, and rough texture, it can use its learned knowledge to recognize that it's likely a type of pear. See in above graph there are only 2 class (0,1) some points lies on zeros and some on one Based on the patterns described in supervised learning, where data is provided with corresponding labels for training, several algorithms are commonly used. Here are some of the key algorithms in supervised learning: Linear Supervised Learning Algorithms:

1. Single Linear Regression
2. Multiple Linear Regression
3. Polynomial Regression

Categorical Supervised Learning Algorithms:

1. Logistic Regression.
2. Support Vector Machines (SVM).
3. Decision Trees.
4. Random Forest
5. Naive Bayes
6. K-Nearest Neighbors (KNN)