Data science used number and name which are also called category and level to predict answer to question, there are basically 5 question data science answers

  1. Is this A or B?
  2. Is this weird?
  3. How much – or – How many?
  4. How is this organized?
  5. What should i do next?

Each of this question is answer by set of machine learning methods called algorithm.

How does data science work?

  • Algorithm = Recipe
  • Your data = Ingredients
  • Computer = Blender
  • Your answer = Smoothie

Is this A or B? (Classification algorithms)

Examples: Will this tier fail in next 1000 miles YES or NO?

Which will bring more customers 5 dolor coupon or 5% discount?

Some times questions can be pharesed in multiple ways with multiple option is this A or B or C or D this is called multi class classification it is useful when you have several 1000 of possibles answers.

Is this weird? (Anomaly detection algorithms)

If you have credit cards already benefited from it, your credit card company analyzes your purchase patterns they can alerts you to the possible fraud. Charges that are weird might be purchased at a store or you are not normally shop or buying unusually pricing item this question is useful in lot of ways for instance you want to know is this presser gauge is reading normal? Anomaly detection algorithms flags unusual events and behaviors and can give you a clue what could be happen for problems.

How much? How many? (Regression algorithms)

These algorithms predicts in numerical such how will be the temperature tomorrow? what will be my 4th quarter sales they answer any question that ask for numbers.

How is this organized? (Clustering Algorithms)

Some time we want to understand structure of a data set. There are number of way to structured the data, clustering is one of the way. In clustering no one right answer.

Examples: which viewer like same type of movies? which printer model fail in the same way?

By understanding how the data is organised you can better understand and predict behaviors on advance.

What should i do now?(Reinforcement Learning Algorithms)

It is inspired by how the brain of a rat and man works responds to a punishment and rewards these algorithms learns from outcomes and decide the next action item. Typically useful for automated system that have to take lots of small decision with out human guidance questions that answer are always about what action to be taken usually by machine or robot.

Examples Temperature control system adjust the temperature up or down or leave where it is. Self driving car whether to break or accelerate. robot vacuum cleaner should keep cleaning or go back to charging station.

These algorithms gathers data as they go. Learning from trail and error.