Machine Learning 101

I always thought Machine learning is a complex term, something very difficult to understand for a basic programmer like me. Because of that belief, I assumed machine learning was not my genre at all.

Today, I regret thinking that way.

Once I started learning it, I realized that the concepts were not that scary at all, as I imagined. In fact, I was able to understand them and even relate them to real-world scenarios. I can now deeply connect the dots. For the first time, I feel like I am part of the crowd that can understand complex topics like AI AND MACHINE LEARNING technologies that are expected to boom in the coming years.

Now, when someone talks about ML, I don't feel lost anymore. I can actually follow the conversation and understand what's going on.

I wanted to share this feeling with my fellow readers, too.

Machine learning is simpler than it sounds. Just like humans learn from experience, machines can also "learn" to give answers. But how does that happen?

With the help of Us Humans

But how?Through Programming.

But how? Using Data.

And that's when it clicked for me.

Machine learning works by using datasets to train and test models, utilizing various algorithms. Through this process, machines learn patterns from data and use them to make predictions or decisions.

Now, we know what machine learning means. There are 3 types of machine learning methods.

1. Supervised learning

2. UnSupervised learning

3. Semi-Supervised learning

Supervised Learning

Supervised Machine Learning is when we know both the input and the expected output. This is the very important point to remember: input and output are already known. And the second point, we are labeling each data into categories, like we can label each item in the dataset and separate them into categories. Like here, for example, we have datatset of animals, and they are labeled as elephant, cow, camel. So, now we know the input (animal dataset), so we are building a model, so that anytime you give any animals to this model, it can easily process and predict the animal names as output.

Sourced from Geeks for Geeks

Unsupervised Machine Learning

Unsupervised Machine Learning is where we know the input, but the output is unknow, what I am trying to tell you is that the machine has to understand the patterns from the dataset, and the machine itself will group the similar data points and will give the output. The dataset will not be labeled. The model has to learn from this unlabeled raw dataset.

Sourced from Geeks of Geeks

Okay, now we understand what supervised and unsupervised learning methods are. But still, we have a few more things we need to dive into.

Supervised Learning Algorithms

We know supervised learning methods take labeled data where the input and output are already known. It follows two types of algorithms to deal with the given dataset

1. Classification

2. Regression

Before we move on and try to understand the algorithm, I want to introduce new terms:

1. Dependent

2. Independent variables

Independent Variable: Here, it meant the input variables. They are considered features. Each input variables are called as features. They are called independent because they do not depend on anyone.

Dependent Variable: Here, it means the output, that is, the target variable, is known as the dependent variable. From the word, we can see that it depends on something. So yes, exactly the output variables completely depend on the independent variables. The model learns to give predictions about the dependent variables based on the patterns it learns from the independent variables.

Regression Algorithm

The regression algorithm is used for continuous variables. It detects only for the features that are numeric values. For Example, the prediction of student marks depending on how many hours the student studied. Here both the student marks and hours studeis is numerical values. Regression algorithm deals only woth numerics.

Some Examples of Regression Algorithms:

1. House Price Prediction

2. Stock Price Prediction

3. Gold Price Prediction

Regression Algorithms

• Linear Regression

• Ridge Regression

• Lasso Regression

• Decision Tree Regression

• Random Forest Regression

• Support Vector Regression

• Bayesian Regression

Now moving on to understanding the next supervised algorithm, Classification.

Classification Algorithm

A classification algorithm is used to predict the output, i.e., target variables. It always says the Yes/No questions. The output will be either yes/no based on the independent variables.

Some Examples of Classification Algorithms:

1. Email Spam Detection

2. Loan Assessment

3. Medical Diagnosis if a person has Cancer or not

Classification Algorithms

• Support Vector Classification

• Logistic Regression

• KNN

• Decision Trees

• Random Forest

• Naive Bayes Classification

Conclusion

Supervised machine learning method used to categorize labeled data where input and output are known. Like inputs and outputs are matched with each other.

Unsupervised machine learning has only input and understands the patterns of the raw data, and builds the model according to the patterns learned.

Supervised learning is task-driven, and unsupervised learning is input-driven.

So, this is just the beginning of machine learning. I am still learning and trying to understand a lot of things. I am trying to understand more about real-time scenarios. Always relate things with realtime scenarios, we will not forget things anytime then.

Happy learning!!