--

# What is structural break?

Structural break is an unexpected change in the pattern of data that we are given to work with.

# Types of structural breaks

1. Cyclic/seasonal — type of structural break where there are repeated patterns in the structural breaks
2. Non Cyclic — type of structural break where there are no repeated patterns in the structural breaks

# How to detect structural breaks?

We can detect structural breaks by looking at the scatter plot if possible or else,

To detect structural breaks, we can use Chow test

Let the model be defined as:

We can add dummy variables as follows

Where Delta t are active at time t with feature X

Now we can do a F test to see whether all dummy variables are 0 or not. If all dummy variables are 0 collectively then we can say there is no structural break otherwise there is structural break

Hypothesis of Chow test

If the null hypothesis is accepted then we can conclude that there is no structural break and the data is perfect to work with

# Ways to resolve structural breaks

1. If the structural break is a non cyclic structural break, we can divide the data into different parts according to the structural breaks and calibrate different models according to different values of X
2. If the structural break is a cyclic structural break, then we can introduce a dummy variable which has a particular value according to a particular period when the structural break is present or it is 0 otherwise

--

--

A data science enthusiast currently pursuing a bachelor's degree in data science