Machine Learning for aspiring Data Scientists :- Part 1

Here in this part of machine learning you will get a brief overview about what exactly is machine learning. In this article more focus has been given on building a path for you to start your journey towards becoming the world No. 1 Data Scientist. You will find brief discussion about the widely used algorithms. You can take this as an initial building block for mastering yourself deep into the concepts of Data Science.

What exactly is Machine Learning ?

It all started in 1950 when Alan turing created the turing test, then in 1952 Arthur Samuel wrote the first computer learning program and in 1967 the neural network algorithm allowing the computers to use pattern recognition. Earlier these techniques were used only to train robots initially. This way earlier the working on these algorithms was not completely autonomous but was mostly manually operated but with time these algorithms have evolved paving its way for deep learning. And now with the enormous number of layers of neural networks even the most complex problems are easy to be solved.

What are the 2 different types of learnings ?

1.   Supervised learning :- So, if you are training your machine learning task for every input with corresponding target, it is called supervised learning, which will be able to provide target for any new input after sufficient training. Your learning algorithm seeks a function from inputs to the respective targets. If the targets are expressed in some classes, it is called classification problem. Alternatively, if the target space is continuous, it is called regression problem.

2.   Unsupervised learning :- Contrary, if you are training your machine learning task only with a set of inputs, it is called unsupervised learning, which       will be able to find the structure or relationships between different inputs. Most important unsupervised learning is clustering, which will create different cluster of inputs and will be able to put any new input in appropriate cluster. Other than clustering, other unsupervised learning techniques are: anomaly detection, Hebbian Learning and learning Latent variable models such as Expectation–maximization algorithm, Method of moments (mean, covariance) and Blind signal separation techniques (Principal component analysis, Independent component analysis, Non-negative matrix factorization, Singular value decomposition) .

 

What are the two different segments of machine learning problems ?

There are basically 2 types of supervised machine learning problems :-

  1. Classification problem :- This is more about classifying the data into the various types. So like a basic example can be if you have data about different fruits and you want to classify them into the different classes then you need a classification algorithm for doing this.
  2. Regression problem :- These problems are the one’s where you have data and you want to get output as continuous value. Similarly a basic example for this problem would be if you have a houses data which happen to be in different localities and you want to predict the prices of some house having similar data related to their dimensions.

There is one subdomain of unsupervised machine learning problem :-

1. Clustering problem :- Clustering is the problem where we tend to aggregate the data onto groups on the bases of similarity patterns in the input data.   This way we tend to solve our complex clustering problems.

Some notable machine learning algorithms ?

1. Linear regression
2. Ridge regression
3. Lasso regression
4. Polynomial regression
5. Logistic regression
6. Support vector machine
7. Dummy regressor
8. Grid Search
9. Random Forests
10.Gradient Boosted Decision Trees
11. Neural Networks

Basic paradigm of how these machine learning algorithms work ?

The Machine learning model is first of all trained and then tested  :-

  1. Train data :- This is used for training your model.
  2. Test data :- This is used for testing your model whether it gives you correct results.

Here we give a basic example of a train which we train and build so that this can be used by millions of passengers later. Next we test the train as to whether it will run or not, which often termed as the testing phase.

Traing the model :-

Testing the above model :-

So once our model a.k.a as train here is trained and tested, we can easily set it operational on the real tracks. Similar to this is our machine learning process where we build up model and also test it so that it works well on the future data. We will continue the machine learning explanation of train and test data in relevance to this example in our next article of this machine learning series.

What online courses to take to master machine learning ?

  1. Machine Learning :- https://www.coursera.org/learn/machine-learning
  2. Applied machine learning in python :- https://www.coursera.org/learn/python-machine-learning
  3. Deep Learning specialization :- https://www.coursera.org/specializations/deep-learning

In the next series of this you will find detailed discussion about the algorithms, starting with the linear regression so as to learn what it is and how to use it for real world problems using python.

Leave a Reply

Your email address will not be published. Required fields are marked *