Thursday, 24 November 2016

Easy Steps to become proficient in Machine Learning:

Easy Steps to become proficient in Machine Learning:

In this post I would like to know you about an easy way of few steps to mastering into machine learning. Before that every one has few basic questions in their mind. Where to start? How to proceed?

Where to start learn ML? 

To understand this post, that is enough to have minimum knowledge on machine learning in Python.The main aim of this post is to know you about the freely available tools to in the market.

I am writing this post by assuming you are not an expert in

1. Machine learning
2. Python
3. Any of Python related machine learning like scientific computing(SCIKIT), or data analysis libraries

Point 1: Need some basic Python Skills

If we go for Python to implement/perform machine learning, we need to have some base understanding of Python is important. By looking at basic tutorial for python is sufficient to get the basic knowledge.

Why we choose python for Machine learning?
Because it is a general purpose programming language, as well as its adoption in scientific computing and machine learning we can go for python.Your level of experience in both Python and programming in general are essential to choosing a starting point.

To full fill this you need to install python in your system, along with that install anaconda it will be useful for scientific computing and machine learning packages at some point. One more reason to install anaconda is. It is an industrial-strength Python implementation for Linux, OSX, and Windows, complete with the required packages for machine learning, including numpy, scikit-learn, and matplotlib.

Point 2: Fundamentals of Machine Learning:
For this we have to answer for once question that “Is it required/necessary to learn kernel methods?”Answer is NO, of course kernel methods help us to understand, create and gain insight from a support vector machine model.for the time being it is not required PhD to write the programs using machine learning, but you need to know the theoretical knowledge of Computer science.

So overall you can get the basic knowledge by browsing the course notes compiled by a former student of the online course's previous incarnation. By using strategic solutions we can move forward towards mastering into the machine learning.

Point 3: Scientific Python Packages Overview 

Well after getting some basic knowledge on python and ML we have to know the packages that are available in python and the use of them. As of my knowledge there are a number of open source libraries generally used to facilitate practical machine learning.

I am listing few of them which are useful at our level

numpy– This package is mainly useful for its N-dimensional array objects
pandas -  This package is used to analysis the data, including structures such as data frames
matplotlib – This is a 2D plotting library producing publication quality figures
scikit-learn - the machine learning algorithms used for data analysis and data mining tasks
Seaborn- which is a data visualization library based on matplotlib.

So while you keep in touch with them it should let you get used to additional and related packages.

Point 4: Start your travel with Machine Learning in Python 

Before start please check once your system contains all the required packages that are listed below
Python, scikit-learn, Numpy, Pandas, Matplotlib. Along with those you have to be good at the basics of the above packages and machine learning basics.

Hurray, your turn has come, start implementing machine learning algorithms with.

Introduction to the scikit-learn.

Point 5: Machine Learning Topics with Python 

The below are the common and useful algorithms that are used with scikit.

k-means clustering: This is one of the most well-known machine learning algorithms.It is a simple and often effective method for solving unsupervised learning problems:

Decision Trees: This algorithm is used for classification

Linear Regressio: This algorithm is used for at continuous numeric prediction

Logistic Regression: This algorithm is leverage regression for classification problems

Support Vector Machines: This algorithm is  a classifier relying on complex transformations of data into higher dimensional space.

Kaggle Titanic Competition: This is a random forest, an ensemble classifier

Dimensionality Reduction: This method for reducing the number of variables being considered in a problem.

Point 6: Deep Learning:

Deep Learning

Deep learning builds on neural network. Let us have a look at a few simple network implementations in Python deep learning libraries

Caffe: Caffe is a deep learning framework made with expression, speed, and modularity in mind.

Theano: This is also one of the Python deep learning librarie. This allows you to define, optimize, and evaluate mathematical expressions involving multi-dimensional arrays efficiently.

No comments: