Time Series Forecasting Version 1.0

Everyone neglect "TIME SERIES FORECASTING",So we decided to come up from scratch version starting from 1.0 to 1.10 on daily Basis.

Read it Out ,come up with your doubt and make it more interesting.

So,let's start with How Important is Time Series Forecasting?

We always neglect Time Series Forecasting but it's is one of the IMportant and Mostly used Concept with ML to solve lot of Projects and to find deep insight from the Data for the better prediction.

As an Data Scientist there are two things ,"To Understand the Data or To Predict The Model."

It's depend on you which path you will choose but as an Data Scientist we should prefer to Understand The Data Before Predicting which is Our True work.

What is Time Series Forecasting?

Time Series is a set of observations on the values that a variables takes at different times.

Time series Forecasting widely used for NON-STATIONARY Data,like economic,weather,stock price etc.

Time Series Data Patterns!!Let

Above Image describing sales Years wise and while have a look on the graph we can easily depict at the start of the Year and end of the Year sale is less.Which in Turn become a Trend known to be as Seasonality(According to Season sales get affect).

As there is presence of "White Noise",so we cannot apply Time Series Forecasting.

Here Question come What is White Noise?

A series purely random in nature is called as "White Noise".

Another Question when "When we come to Know there is presence of White Noise"?

When Mean of Series is equal to Zero and Correlation also tend to zero.

We cannot find purely white Noise,we say it is near to White Noise.

AutoRegressive(AR) Model?

As name implies we Regressed itself with a timeline.

Let's say we have Sales Data of two years and we have to predict Sales in the Month Of June?

So,How AutoRegressive will work?

Here,Yt depends only of Past Values.Yt-1,Yt-2,Yt-3 etc.

So,general equation will be Yt=Bo +B1*Yt-1

Optimizing Above Equation!

Yjune=Bo + B1 * Ymay (Bo is an intercept,only Y in the form of Y=ax +b)

Here P=1 and if P=2,then we have to take last two months to predict the required Sales.

Let's Predict sales in the Month Of July!!

Yt=Bo + B1 *Yt-1 +B2 *Yt-2

Optimizing above equation

Yjuly=Bo + B1 *Yjune + B2 * Ymay

If correlation between Current and Past value is High then AutoRegressive works better.

Hope this will Help you with Basic and tomorrow Second version will be Posted.

Be Updated!!

Happy Learning!!