Machine Learning in Nutshell:

Machine Learning is defined as Finding Patterns in the data by using Algorithms and making the model then we enter new data to see if it matches the pattern.  


Requirements:

  1. Lots Of Data 
  2. Lots of Computing Power 
  3. Effective Machine Learning Algorithms 

These three things are now more and more available.

Who are interested in Machine Learning? 

  1. Business Leaders (who want solutions of their business problems)
  2. Software Developers 
  3. Data Scientists( want powerful, easy to use tools)

Who are Data Scientists? 

A data scientist uses data to understand and explain the phenomena around them, and help organizations make better decisions.

True Data Scientists are scarce( not easy to find), Very Expensive 💸.

Ethics of Machine Learning:

Your data Should not be Biased (unfairly prejudiced). Otherwise You will end up with false Model. 

Main Points:

  1. Machine Learning let us find patterns in existing data, then creates a model and use this model to recognize those patterns in new data. 
  2. It has gone mainstream. 
  3. Although it raises ethical Concerns. 

It can help your organization to grow rapidly.

Machine Learning Process:

  1. Iterative (both big and small ways)
  2. Challenging
  3. Often rewarding (but not Always) 

Ask the right Question:

Choosing what question to ask is the most important part of the process.

You should ask your self did you ask right Question or Data? How will you measure Success? 

Machine Learning is an Iterative Process:

You first Apply Preprocessing On the Raw data because your data will be having missing, extra and repeated data and convert it in to Prepared data( We spend most of the time in this process because prepared data is the key to make useful model.)


We then create the candidate model. This model obviously not 100% ready. So you also iterate this Process to get the refined model. We then iterates the whole Process time to time to update it. Because world is changing every second 🙂.


A closer look at Machine Learning:

  1. Training Data 
  2. Supervised Learning(When data is labelled. The value you want to predict is in training Data.)
  3. Unsupervised Learning(The data is not labelled. The value you want to predict is not in training Data.)
  4. Classifying Machine Learning Problems and  Algorithms
  5. Training a Model 
  6. Testing a Model
  7. Using a Model 

Implementation of Machine Learning:

  1. Create custom models in R and Python using general ML packages.
  2. Creates custom models using focused packages, e.g., Tensor Flow etc.
  3. Create custom models using cloud services. 
  4. Use pre-defined models, e.g., Azure Cognitive Services.

Summary:

  1. Machine Learning has come to age. 
  2. it is not hard to understand. 
  3. It can be hard to understand. 
  4. It is time taking process.
  5. It can be helpful for your Organization. 

Hope you are able to get the concept of Machine Learning.

If you have any queries do let me know in comments. If not, you can say hi too. I love it 🙂.