We learnt the defination of machine learning and it's conceptual overview in previous post. We wil now move on to the components of machine learning. Components for machine learning are represented in different forms from different source. I will be disscusing the basic components that constitute the Machine learning approach.

1. Data:
   I don't have to say much about data, right? We expirence it in each moment of life. The blog you are reading, Tea/coffee in your hand, thoughts inside your brain, even counter arguments you are having against this post all represents common information of our daily life. These are the one form of data, may be these examples  are not intresting for you. let's pick up the unusaul one, your heart beat. Does it tell something about your health? or about quality of life? or your physical activity? let's assume we have record of your heart beat 

   
   
   image source: vowe.net

   We can predict some physical activity going on for higher heart rate and indivdual having proper rest with lower heart rate. Also, if you are professional for health practise this data can give an idea of mental stress, fear or excitement.
   
   We have disscussed the general example of what could be the data. There are lots of possibities depending upon your power of imagination. The Credit/Debit card you have, message you type on social media, places you travel or anything else that intrest you are good examples of data that could be studied. For geospatial comunity, the satellite data that we have for more 40 years, GPS trackers of your mobile or watch or any trackers and other spatial sectors you can name it.
   I don't want to go on characteristics of data that should be considered on this post may be i would dedicate seperate post for it. For now we will go on general overview only.

2. Model:
   Model is simply the guiding principle on which data and results interacts with each other. For simplicity we take an example of linear regression, where interaction between input and output varaibales is defined by best fitted simple linear equation. But the derivation of these equation is governed by the learning process which will be disscussed below. For now, we will observe the graphs for simple linear models.
   
   I have taken random simple graph representing linear model. let's for  example x-axis represents the number of sales of product and y-axis represents the profit earned by the company from sales. With this graph we can observe there is sort of linear relation in between these variable. And we have to be clear that there is no perfect representation of models, only thing we will try to do is search the model with less discripancy or in other words we search for models that nearly represents the reality of nature.
    
3. Learning process:
   Learing process is guided by mainly 2 principle objective function and optimization. Objective function is problem driven. For instance we have assumed linear regression eaquation as the relationship for number of sales and profit earned in above example, right? let's take an another example if we observe number of days and covid infections now our objective function won't be linear regression, rather it would be exponential function. This is the way we think for defining the objective function for the model. This fucntion gets more complicated upon the depth we go for data analysis. I don't want to dive deep into this for now.
   Then what about the optimization? Did you get the objective here? Again from above figure, we can clearly see the linear equation don't exactly fit to our data and it never does in real world problem. Only thing we can do is minimize this mis fit and try to go towards the reality. This is the optimization  technique used in machine learning. It plays a central role in machine learning, as almost all machine learning algorithms/modals use optimization to fit a model to our dataset.
    
4. Application:
   Now you must have get the point, we have wider range of application with machine learning. It can be from profit projection for normal office to national level growth projection on various aspects. There are no limits and boundary for the apllications, that's why various startsup have arised using machine learning and AI.