3 Advantages to Time Series Analysis and Forecasting
A time series is a collection of observations in chronological order. These could be daily stock closing prices, weekly inventory figures, annual sales, or countless other things.
Time series analysis, then, is nothing more than analyzing—plotting, identifying patterns, etc.—the series.
Finally, a time series forecast is taking those past observations and making predictions about what will happen in the future if the same patterns continue to hold true.
Time series analysis and forecasting are among the most common quantitative techniques employed by businesses and researchers today. We will dive deeper into the three major advantages of performing time series analysis.
1. Time Series Analysis Helps You Identify Patterns
Memories are fragile and prone to error. You may think that your sales peak before Christmas and hit their bottom in February… but do they really?
The simplest and, in most cases, the most effective form of time series analysis is to simply plot the data on a line chart. With this step, there will no longer be any doubts as to whether or not sales truly peak before Christmas and dip in February.
Now that we have plotted our sales data, it becomes immediately clear what patterns we have. Sales are trending upwards year-over-year, and seem to follow a regularly yearly pattern. The months of January and February see the lowest sales figures and there is a major spike in November and December.
2. Time Series Analysis Creates the Opportunity to Clean Your Data
In the example above, we plotted actual sales figures for each month in the data set. If any observations were missing, the gap in the time series chart would show that right away. With any gaps in the data identified, it would be easy to impute those missing values (that is, fill in the gaps with some calculated value).
Furthermore, we would be able to identify outliers in the data. Perhaps instead of looking at actual sales, it would make more sense to plot the percentage difference between observations. This is a technique that can help smooth out very noisy data.
In this case, by plotting the percentage difference in sales from month to month we smoothed out much of the data—except for the enormous spike in March 2017. This is not necessarily a bad thing, however; without performing these analytic steps we may have been unaware that such a spike existed.
3. Time Series Forecasting Can Predict the Future
If we could look into a crystal ball to see the future, we would all be rich. Knowing when to expect a lull in sales, a slowdown in inventory levels, or a surge in demand would be incredibly valuable for any company.
Although not a crystal ball, time series forecasting methods can help us gain a useful glimpse of the future. While mathematically dense, the thrust of forecasting comes down to looking at past behavior and extending those patterns into the future.
In the example above, the forecast is given by the blue line and band. The forecast clearly continues the upward sales trend and even exhibits the seasonal dips and spikes that we have come to expect.
While you should not treat such a simple forecast as gospel, it can be a helpful indicator of what to expect in the near future. With this information in hand, you can more easily prepare for the times ahead.
If time series analysis and forecasting has piqued your interest, consider registering for my in-depth webinar on Wednesday, March 10th 2021 at 1pm CST.
During the 1-hour live stream, I will teach you step-by-step how to build a time series forecast model in Alteryx. You will learn:
- How to take advantage of the best Time Series tools in Alteryx
- The difference between ARIMA and ETS time series models
- How to identify patterns found in your historical and transactional data
- How to account for seasonality in your time series forecasting
- How to identify risks as well as opportunities based on those patterns
If you’re interested, register here for my “How to Create a Time Series Forecast in Alterxy” webinar.
All attendees will receive a copy of the workflow and data set used in live stream, plus they’ll get a link afterwards to the recroding.