Time Series Analysis
Time Series Analysis in Big Data Analytics
- It is series of notes of categorical / numerical variables indexed by a date / timestamp
- The underlying example is a time sequence of prices of stock, it represents the general structure of the data time series and is noted in a hourly basis
- The first process in the analysis of time series is plotting the series, with the help of line chart
- Its main purpose is forecasting future values by using temporal format of a data
- e. forecasting values from future
- To construct robust prediction, temporal ordering of data is required
- The commonly employed model for time sequence analysis is ARMA (Auto Regressive Moving Average). It has 2 parts which includes auto regressive and moving average part
- It is termed as ARMA(p,q) model where p is auto regressive part and q is moving average part
Auto regressive Model
- Its short form is AR(p) .It is studied as auto regressive model of p order.
- Its mathematical model is represented as
- φ1, φ2, … φp are arguments to be determined
- c is constant and
- εt denoted the noise
- This model is not in motion as a few conditions are required on the values of arguments
Moving Average
- The short form MA(q) denotes the moving average model of q order
- Its mathematical form is denoted as
- θ1, θ2, … θq are the arguments of the model
- μ is the expectation of Xt and
- εt, εt-1 are error terms of noise
Auto regressive Moving Average
- this model merges auto regressive and moving average model
- It is shortly noted as ARMA(p,q)
- The mathematical format of its representation is
- It is a blend of AR(p) and MA(q) models
- The underlying source code portion shows the way to deploy an ARMA(p,q) in the programming language R
- Data plotting is the initial process to determine any temporal format in the data
- It is clear from the plot that there is powerful spike at the last part of every year
- The underlying source code fixes the ARMA model to the data
- It executes various mixture of models and chooses the one which has minimum error
.
