Saturday, June 23, 2018

Introduction to Machine Learning

 What is Machine Learning?


           Machine Learning is one of the many approaches to Artificial Intelligence (AI). It provides systems the ability to learn and improve from experience. The machine learning programs work on the data provided and try learning by themselves. The basis of machine learning is to build algorithms that can receive input data and use statistical analysis to predict an output while updating outputs as new data becomes available. There are different types of machine learning models which will be covered later. 
         


Why Machine Learning?


         It is not possible for any human to go through enormous amount of data, say 500,000 patient records, let alone drawing inference from it. The data, in this case - patient's records, may have different features from age, RBC/WBC count, symptoms, et cetera. Considering all these features and trying to draw an inference seems superhuman task.
       
       Now, what are computers capable of? Processing information and calculations at a very high rate is what a computer does. Using such processing speed to draw inference from a large set of records seems now possible. It can even analyse complex data. With increasing volumes and different varieties of data in past few years, there is a need to draw inference from this data to solve real world problems.
                               


Machine Learning Process:


                        


          The learning process has different stages. First is the data pre-processing step which is very important. The raw data provided, sometimes, can have missing, unnecessary elements in it. Before providing the data to the learning model, these things should be taken care of. Also, in some cases, there will be a need to normalize, scale the data, only after which the data is given to the machine learning model.

          Secondly, visualize the data. It gives more insight in the training data provided, which helps in further modelling. The patterns within the data can be easily detected by visualization.
    
          Having visualized the data with a slight hint of which model to be using, choose the right model and provided the structured data to it. The learning model will then be able to learn from the data and will have the ability to predict the outcomes for some new data.

         Different algorithms have different parameters associated with it. Tuning in these parameters helps you get a better efficiency of the learning model. All these things can be done using different libraries provided by the platform on which you run the program. 
     

Applications of Machine Learning:

              The applications of machine learning are endless. It can be used in diverse fields like weather prediction, medical analysis, email Spam classification, detection of text from images, recommendations in online shopping, booking, et cetera. A few links to some successful machine learning applications are:

Image Sources:
  1. Image
  2. Image
  3. Image

Random Forest Classifier

                 Random Forest Classifier is a  supervised   classification  algorithm. It is very popular and performs better than most c...