Feature Scaling - Subtract Mean and Divide by Standard Deviation

Supposing we are comparing two different features like Fahrenheit and Celsius. In such a case, we cannot directly compare these two features and arrive at a statistical judgement because, they both have diffrent measures of unit.

Another example is that of predicting the price of the house based on features like number of bedroom, vicinity to school/hospital, number of floors, total area in square feet and age of the house. On this case too, we cannot directly compare the features of the house as they all have different units of measurement; also they possess different range of values.

Hence before comparing them, we must first remove the bias of scaling unit as well as bring them to a similar range of values.

Feature scaling is therefore a method used to standardize the range of independent variables or features of data. 

 Supposing our feature is X, the normalized form of X would be:

                             X_Norm=(X-mean(X)) / StandardDeviation(X)

For those of you who do not have a strong background in Quants and Stats, I have tried to explain in simpler terms. Hope this helps!