Free Business Intelligence Practice Tests - Data Analytics Test
Techniques used
Till now we had discussed what “Data Analytics” is and what the types of “Data Analytics” are. But now let’s understand the techniques used for “Data Analytics”. The most commonly and widely used technique for “Data analytics” is “Statistical Methods”. The other methods which also contribute in “Data Analytics” methods are Neural Network, Statistics & Fuzzy Logic, & “Genetic Algorithm”. Let’s discuss Neural Network, Statistics & Fuzzy Logic one by one.
Neural Network
Neural Network is one of widely used technology for the analysis of data. A neural network works similar to the human brain which is composed of neurons. Its name is derived from the same concept. Basically a neural network is based on small chunks known as neurons, which combine together to form the complete neural network. The logic of neural network is defined through these neurons. Some neurons are only composed of weights to form a simple neural network while some are composed of complex calculations to form neural network. The outcome of neural network is decided by these weights or calculations.
For example, a neural network might be used for handwriting recognition, or for voice recognition. Sometime neural network recognition techniques used only partial match method which might be impossible with algorithms. The inspiration behind for neural network is human brain.
As from the figure it is clear that, a human body neuron takes several inputs from brain, combine them together and transfer that message to different body part. Similarly a neural network takes several inputs, calculates their weights, combine them together again perform summation operation with new weights and finally comes out with an output.
Statistics
As the name suggest, the Statistical Methods has to do something with statistic. In actual it deals with the meaningful quantities of events, objects, persons etc. This method is very popular with “social science data analysts”. The statistical method de-synthesizes data, information or any factual object or element to answer the research query. It’s approach it to use a systematic process to answer the research query. The statistical method has two main approaches
- Descriptive statistics, and
- Inferential statistics
Descriptive Statistics
In this approach, we make a referral master data. Using this master data we may compare, examine or explain the abstraction of phenomena’s. For example if X then Output will be Y. Whenever a problem is descriptive in nature we refer to Descriptive Statistics.
Inferential Statistics
In this approach, we use master data to conclude the abstraction of phenomena of some popular parameters. For example: Y = F(x). Whenever a problem attempt to infer, influence, predict, cause-effect or include relationship we may use Inferential Statistics.
Fuzzy Logic
Fuzzy Logic is similar to probabilistic logic. It is based on approximation method. Fuzzy logic is the extension of partial truth method. In binary set, it can have only two possible values, i.e. 0 or 1 or in simple word true or false. But in fuzzy logic, variables may have value range in between 0 to 1 in degree. This means the fuzzy logic variable range from complete truth to complete false.
Mathematically both probabilistic logic and fuzzy logic are same, but conceptually fuzzy logic define the “degree of truth” where as the probabilistic logic define the “probability of occurrence”.
The fuzzy logic usually implements the “IF-THEN” rules. The Fuzzy Logic rules say “IF variable IS property THEN action”. There is no “ELSE” rule in fuzzy logic.
For example: “IF water IS overflowing THEN switch off water tap”.
