A knowledge of statistics is like a knowledge of foreign languages or of algebra; it may prove of use at any time under any circumstances by A.L. Bowley, Statistician and Economist.
Uni-variate Analysis
As the name Uni-Variate itself suggest the meaning i.e. one variable. It is the simplest form of data analysis. It only accounts for one variable's effect on a dependent variable. It does not deal with any causes and relationships because there is only one variable, you can not compare it with other variable.
It explores the variable one by one. Variable can be either categorical or numerical. It only provides the summary of the variable and any other specific pattern associated with it. It is a common method for understanding the data.
 |
| Frequency Distribution Data |
 |
| Datacian |
This is one of the way to understand uni variate data. You have options of Bar charts, histogram, frequency tables etc. And also you can use mean, median, mode, range, variance, maximum, minimum, quartiles and standard deviation to understand more about the data.
Bi-Variate Analysis
Bi variate is the simultaneous analysis of two variables(features). It explores the relationship between the two variables, like what kind of association they have, or how they are correlated with each other and their strengths.
It can be done between numerical & numerical variables, categorical & categorical variables, numerical & categorical variables. There are number of graphs or techniques are available to play with the bi variate data like bar graph, scatter plot, whisker plot etc. You can so regression analysis or can check the relationship through the correlation.
 |
| One more column added: Sales(Week wise) |
 |
| Datacian |
Multi Variate Analysis
Multi variate analysis comes into picture when you have to deal with more of complex data. In real life, our target variable is caused, or affected by many other different kind of variables.
When you have to deal with more than two variables at once, have to do multi variate analysis. There are many types of techniques that are involved in this type of analysis, some of them listed below.
1. Cluster Analysis
2. Principal components
3. Correspondence Analysis
4. Factor Analysis
5. Multi dimensional Scaling
6. MANOVA
 |
| Source: Colorado.edu |
The graph shows the clustering results by k-means clustering algorithm.
Feel free to comment your opinions!
0 Comments