{"id":12499,"date":"2020-02-20T12:22:42","date_gmt":"2020-02-20T06:52:42","guid":{"rendered":"https:\/\/www.mygreatlearning.com\/blog\/time-series-analysis-and-forecasting\/"},"modified":"2024-09-03T15:01:37","modified_gmt":"2024-09-03T09:31:37","slug":"time-series-analysis-and-forecasting","status":"publish","type":"post","link":"https:\/\/www.mygreatlearning.com\/blog\/time-series-analysis-and-forecasting\/","title":{"rendered":"Learn to Forecast a Time Series in Python - All You Want to Know About Time Series Analysis"},"content":{"rendered":"\n

Time series analysis has been widely used for many purposes, but<\/gwmw> it is often neglected in machine learning. A time series can be any series of data that depicts the events that happened during a particular time period. This type of data often gives us a chance to predict future events by looking back into the past events. Nevertheless, it is also interesting to see that many industries use time series forecasting<\/a> to solve various business problems. Before diving deep into the application of time series, we need to see the various parameters which should be considered in forecasting.<\/p>\n\n\n\n

Parameters<\/strong><\/gwmw><\/h2>\n\n\n\n

Stationarity<\/span><\/gwmw><\/h3>\n\n\n\n

Data attains stationarity when the properties such as average value, variation of the data, and standard deviation do not change over time. It is also an assumption to hold as the data which is not stationary can wrongly forecast results. Here\u2019s an example:<\/p>\n\n\n\n

If we look closely into the above diagrams, the left diagram clearly tells that the data has an upward trend but it is not stationary. The graph clearly shows that the mean over the different time periods is not the same and is rapidly fluctuating. On the other hand, if we look into the right one, the data has no trend. It moves over a straight line and has a constant mean and variance over different time periods. Hence, this type of data is suitable for analysis.<\/p>\n\n\n\n

<\/p>\n\n\n\n

Seasonality<\/span><\/gwmw><\/h3>\n\n\n\n

Data is said to be seasonal when it changes exactly in the same way over a different time period. Let us consider the following data.<\/p>\n\n\n\n

If we look into the data, it contains the death toll of people from 2006 to 2016 i.e 10 years. Now we can see that the upward fluctuations occurred during the starting of every year and it came to a minimum during the mid-year period. This clearly implies that the death of these people follows a pattern and it can also be inferred that the death after the mid-year period is higher than the initial 6 months in every year. This is considered as another assumption of a time series data which needs to be true for successful forecasting.<\/p>\n\n\n\n

Autocorrelation<\/a><\/span><\/h3>\n\n\n\n

This is another important parameter which assumes that the data needs to be auto-correlated. Data is said to be auto-correlated when it shows similar observations over a time lag present in the data.<\/p>\n\n\n\n

Let\u2019s take a closer look at the chart above. If we see the data, we see that the first value and the 24th<\/sup> value happen to be the same. Also, the 12th<\/sup> value and the 36th<\/sup> value happen to be the same. A quick inference? The graph implies that the data follows a high correlation as it is showing similar observations in every 24-hour period. Although it is quite difficult to get data with these types of correlation, we can hold an assumption that the data should be approximately correlated.<\/p>\n\n\n\n

Take a look at this free resource for you to enhance your knowledge of time series: Time Series Analysis Stock Market Prediction Python<\/a>.<\/p>\n\n\n\n

Applications<\/strong><\/gwmw><\/h2>\n\n\n\n

Time series analysis can apply to many sectors including but not limited to:<\/gwmw><\/p>\n\n\n\n

    \n
  1. Power generation companies to observe the usage of electricity <\/li>\n\n\n\n
  2. Financial sectors such as banking, stock market, etc.<\/li>\n\n\n\n
  3. Weather forecasting to predict the weather.<\/li>\n\n\n\n
  4. Traffic engineering to study traffic volume during peak hours.<\/li>\n\n\n\n
  5. Forecasting hospital blood requirements<\/a><\/li>\n<\/ol>\n\n\n\n
    \n