Time Series Analysis
Description
Time Series can be simply seen as a stochastic process obtained from real data. The stock index is a good example of a time series as its value changes with uncertainty over time. Time series analysis aims to understand and predict the movement of a dependent variable over time.
The biggest difference between time series analysis and regression analysis is that, while regression analysis assumes that the independent variables are independent from each other and have their own independence, time series analysis assumes that variables have autocorrelation. Also, regression analysis does not care about the order of data, whereas time series considers the influence of previous data on subsequent data important for the analysis.
Stocks are also a good example of autocorrelation. In the KOSDAQ market, whether the face value of a stock will rise or fall cannot be known, but if it is 10,000 won per share today, it could go up to at most 13,000 won or drop to 7,000 won tomorrow. Of course, this is because current laws do not allow for fluctuations of more than 30%p, but the face value of tomorrow $Y_{t+1}$ is definitely dependent on the face value of today $y_{t} = 10000$. Thus, even though it fluctuates randomly, the goal of time series analysis is to catch changes that are somewhat correlated with the current data. [ NOTE: The face value of tomorrow is represented as a capital letter because it is still unknown and thus considered a random variable, while the face value of today is represented as a lowercase letter because it is already known data. ]