Time Series Analysis
In this series I am going to provide you very brief introduction about time series analysis.
Introduction
- Time play a very important role in success of any business, here time series helps us to see ahead of time.
- Time series is a sequence of equally distributed data point in order of time. Data point can be presented on List or a chart.
- This is a statistical technique for trend analysis
Time Series Analysis?
- Time Series analysis is a way to analyzed time data to get important statistics and other useful points of time series data.
Time Series Forecasting?
- Time series forecasting is use of a time series model to predict future values based on past observed value in time series.
Applications
- Stock Market Analysis
- Economic Forecasting
- Sales Forecasting
- Financial Market Analysis
- Inventory Management
- Yield Projections
- Military Planning
- Marketing and Sales Forecasting
Lets explore some basic terms used in time series.
Time Series
- Machine Learning dataset contain a collection of many attributes or observation.
- Example
- Attribute
- Attribute
- In time series dataset a time attribute is added. As Time play a very important role to predict value of future dates.Example
- Time, Attribute
- Time, Attribute
Time series components
Trend
It is a increase or decrease of behavior of data over a period of time. It can be linear or non-leaner. If there is upward or increase behavior called as Up-Trend, same for decrease know as Down-Trend. When there is no trend that is known as stationary trend or horizontal trend.Trend appear for some time and disappear.
- Example :-A packers and movers open new office at newly constructed society. That will get good business over a period of time like 6 Month until all houses get filled. After that it will be no or very less business.
Seasonality
- It is very much similar to trend but in repetitive in nature.
- It is also increase and decrease in behavior data over a period of time.
- Example : – sales increase each year on black Friday. Gifts purchases increase each year on Christmas and other festivals and it happen each year.
Noise/Irregularity
- Variability in data that cannot be explained by model. It also called as residual.
- It happen for a small amount of time and non-repeatable mode.
- Example :- earthquake in a city can increase demand of medicine and some other day to day living stuffs. But it will end once everything is fine. It happen once and not predicable.
CYCLIC
- It is known or describe as a non-fixed pattern for more than a year amount of time.
- It can happen any time like 3 years, 5 year and keep repeating and hard to predict.
When this not applicable
- When data is constant that time then this is not applicable.
- When data is in form of a functions, where we can put function to get result.
I am going to cover in-depth concept and particle aspect in coming posts.
