In our last videos we had completed with logistic regression and its practical session in this video we will be starting with time series forecasting. So first we'll do what is the concept of time series time series analysis is based on the assumption that successive values in the data file represent consecutive measurements taken an equally spaced time intervals, there are two main goals of time series first identifying the nature of the phenomena represented by the sequence of observations then then the next school is the forecasting that is predicting future values of the time series variable based on the present data. Both of these goals require that the fact you know observed time series data is identified and more or less formally described. Once the pattern is established, then we can interpret in integrated with other things. So, time series analysis comprises methods for analyzing time series data in order to extract meaningful statistics and other characteristics of the data time series forecasting is the use of a model to predict future values based on previously observed values does by time series, we mean numerical determination which are collected, observed or recorded at successive intervals here as represented a time series data graphically, the use of time series models is twofold.

First, obtain an understanding of the underlying forces and structure that produce the observed data next, fit the model and proceed to forecasting monitoring or even feedback and feed forward control. Now let's do a comparative study between time series data cross sectional data and panel data in time series data it is a collection of observations or behaviors for a single subject or entity at different time intervals that is it is generally equally spaced. For example, suppose We are collecting maximum temperature, humidity and winds all these three behaviors in New York City, New York City is a single entity collected on first day of every year that is multiple intervals of time. So, this is time series data. Next, what is cross sectional data it is a collection of observations behavior for multiple subjects or entities a single point of time. So, in cross sectional data, we have multiple behaviors and multiple entities a single point of time.

So, for example, we are welcome collecting maximum temperature the data for maximum temperature, humidity and wind that is these are multiple behaviors data for multiple entities that is New York City SFO Boston Chicago, that is multiple entities and it is collected for a single point of time that is in first gen 2015 that is 112 thousand 15 that is single instance. So, this is cross sectional data next what is panel data panel data is usually called as cross sectional time series data that is it is combination of both. It is a collection of observations for multiple subjects at multiple instances example max temperature humidity and wind that is multiple behaviors which is collected from multiple entities that is in New York, SFO, Boston, Chicago and first year of every year that is multiple intervals of time. So, in panel data, the behaviors are also multiple the entities are also multiple and the current time intervals are also multiple.

So these are the differences between the basic differences between time series data cross sectional data and panel data. Next now let's come to their different applications of time series analysis. First application is stock market analysis like we can predict the stock market prices based on the present rises we could predict the future stock market prices weather forecasting, suppose you want to predict the amount of rainfall that will happen in the future. So in the present based on the present rainfall rates we can predict the future in for calculation of GDP based on the present GDP future GDP can be predicted focusing exchange rates. This is very important based on the present exchange rates of our economic we can forecast the future exchange rates that is based on the present macroeconomic conditions and based on the present exchange rates, we can focus Future exchange rates scales forecasting based on the present sales of different products we can focus on future sales next fashion industry that is we can forecast the future trends that is that are going to come in fashion industry based based on the present trends, then economic forecasting, economic forecasting is forecasting the future condition of economic using different economic indicators like future inflation rate, future unemployment level, future price level future GDP, etc budgetary analysis we can predict the future budget based on the present budget then census analysis like suppose if we want to predict the future population level of our country, we can use the present population level and predict the future population level.

These are the different applications of time series analysis that we have discussed. Now, let's come to the different assumptions of time series analysis. The first and foremost important assumption is time series analysis assumes that whatever has happened in the past will continue to happen in the future. And the next assumption is time series analysis cannot predict random shocks. Like for example, if there is a Brexit or C An earthquake or there is a flat or there is a drop or there is a for me these random shocks or irregular component cannot be captured by time series analysis. So in this video we will be learning till here before I move to the next video let me recap the concepts that we have covered in this video.

In this video we have discussed the concept of time series where time series analysis is based on the assumption that successive values in the data file represent consecutive measurements taken and equally spaced time intervals, we have done a comparative study between the time series data cross sectional data and panel data that is in time series data the behaviors are multiple then there is a single entity at multiple intervals of time for cross sectional data, there are multiple behaviors there are multiple entities a single point of time and panel it is combination of time series data as well as cross sectional data. That is there are multiple behaviors, there are multiple entities and at multiple intervals intervals of time. Then we are discussed the different applications of time series analysis, that is stock market analysis, then tenses analysis predicting fashion trends in fashion industry, economic forecasting, GDP forecasting, sales forecasting, forecasting future exchange rates, budgetary analysis and etc.

Then we had discussed the different assumptions of time series analysis that is time series analysis considers whatever has happened in the past will continue to happen in the future and time series analysis cannot predict random shocks that is Brexit or say any irregular component like earthquakes, floods, droughts, fermions etc. Now, in our upcoming video, we will be discussing about the different components of time series analysis that is secular trend or trend component cyclical trend or because a cyclical component that we'll discuss about the seasonal component of time series, and then we'll be discussing about the random components or the irregular components of time series. So for now, let's stop this video over here. Goodbye. Thank you. see you for the next video.