Data Science - Time Series Analysis
It is a set of observation or data point s which are taken at specified time period. It’s said that the most effective way of time series is to maintain the equal intervals of time to calculate the correct prediction. Business forecasting is a part of time series analysis, stock market works on the prediction model. While determining the past experiences and scenarios one can invest in share and predict how things will turn up in future. Also a lot if retailers make bulk purchases on the basis of predicting the sales the will achieve in forthcoming future. All this is part of business analysis, which is a part of almost all the domain irrelevant of their nature. Other major terms are analyzing past behaviour, future plans and evaluation of current accomplishments.
Past behaviour is nothing but patterns that are being observed in the past, the season of sale, product preference etc. all this comes under past behaviours. Future plans now once you know the past behaviour it’s very easy plan the future investment, stocking and spending. And finally evaluating the goal that has already been achieved, every professional work on goal basis and with help of Time Analysis new goals can be achieved.
So, let’s suppose there is coffee shop owner, after a successful sale in first few months. How is going to calculate the sale? He will sum up the number of servings in those months right? But what if he wants to predict the sales of the coming month, and you just have two variables for it that is time and sales of previous months. Here Time series analysis comes into the picture and this is where is has been used to forecast the coming opportunities and warnings.
Components of Time Series Analysis
Trend- The three sorts of trends are Up trend, Low trend and horizontal trend. Let me put an example of trend, so there is new township opened and someone started a hardware shop in there. Now what will happen, the people who are going to accommodate there they will buy stuff from that shop and the sales of that hardware shop will go up and time series will show up trend. Once every house is settled will then there will be low sale, showing down trend. And once the trend graph will not go up and down but stay static will become horizontal trend. Trend is something that happens for some time and then it disappears.
Seasonality- A repeating pattern with is a fix time period, just like every year the business of sweets rises up in the festival season. This repeating pattern doesn’t change but repeat the same business on seasonal basis.
Irregularity- This component is best defined as the- let’s suppose if any calamity happens the sale of particular medicine or ointment increase, which is erratic and once the people are healed then again the sales will be back on its pace. So Irregularity happens this way and affects the time analysis and number of sales can’t be measured.
Cyclic- No fix pattern, keep on repeating a very tough to predict and repeat up and down movement.
Data Science includes this major topic which is time analysis. It is beneficial in many ways and is one of foundation algorithm which is applied by all data scientists.
