Data Analytics: Linear Programming

Managers can optimize their resources by using Linear Programming(LP).
All Linear Programming problems must have the following 5 characteristics:
i) Objective Function
ii) Constraints
iii) Non-negativity
iv) Linearity
v) Finite
The following 2 assumptions are made:
i) Resources are limited
ii) There is only one objective

Two ways to solve Linear Programming Problems:
Correlation and Regression

Correlation tell us about the direction and strength of the relationship between variables.
Direction is positive or negative while strength is strong or weak. In correlation, direction is more important than strength.
For quantitative variables, we use Pearson Product Moment Correlation and for variables that are qualitative in nature: Spearma Rank Correlation.

Regression shows the relationship between two or more variables and also the cause and effect. This can be used to forecast or predict. In all Linear Programming problems, the maximisation or minimisation of some quantity is the objective. All Linear Programming problems also have a second property, restrictions or constraints.

Assumptions for Linear Regression:

• Linear relationship
• Number of sample size should be more than number of parameters
• Normal Distribution
• Multicolinearity
• Auto correlation