Fall has become the most imminent public health problem in the world. Each year thousands of people over age 65 experience falls. Because of that a large number of elderly people gets injured. Even some of them die if not rescued on time. With such immense increase in the falls that harms the individuals physically and/or mentally Researchers found ways to tackle the problem by devising different techniques for the purpose of fall detection. Most of techniques being used rely on Threshold based algorithms (TBA). While some of them brought the Machine Learning (ML) into the field to fight detect falls. The use of TBA and ML combinedly for fall detection has been reported. But still every technique has its advantages and disadvantages. We are doing analysis of techniques that are totally based on either TBA's, ML or the combination of both TBA and ML.