James has an MBA and a MA in Humanities. He writes on leadership, business strategy and finance.
In this lesson, we will explain time series analysis, its purpose and uses. We will define trend, cyclical and seasonal variances. We will solve a real-world business problem using time series analysis.
Predicting the Future
NRAR Robotics recently introduced a new line of personal robots to the consumer market. After about three months of slow sales, the sales of robots had increased to an average of 10% for the next nine months. It takes the company about three months to build a new robot. The management team needs to forecast the demand for three months in advance to ensure enough robots are available. Time series analysis helps the team improve the forecast.
Time Series Analysis
Time series analysis is the collection of data at specific intervals over a period of time, with the purpose of identifying trends, cycles, and seasonal variances to aid in the forecasting of a future event. Data is any observed outcome that's measurable. Unlike in statistical sampling, in time series analysis, data must be measured over time at consistent intervals to identify patterns that form trends, cycles, and seasonal variances. Measurements at random intervals lose the ability to predict future events.
Trends are consecutive increases or decreases in a measurement over time. A trend could last several days, months, or years. In almost every observation, a trend will reverse itself during the lifetime of measurement. This reversal is sometimes referred to as a correction. Corrections occur in the economy, in a stock market, and in a business. It normally follows unprecedented growth or loss. A cycle, on the other hand, is a pattern of growth, followed by a decline, followed by growth. The cycle is recognized by the pattern of a rise and fall repeated over several periods of measurement.
Seasonal variances are measured over several months and are associated with a specific time of the year. Retailers realize a seasonal growth in sales during the months of November and December. For the rest of the year, sales are relatively flat. The year can be divided into four quarters. The first three quarters show a small volume of sales giving way to a large growth in sales in the fourth quarter.
The insights discovered during a time series analysis, such as the spike in sales for the retailer, can be used to forecast demand for subsequent business periods. In fact, many retailers place purchase orders for the holiday shopping season months or even a year in advance based on these forecasts.
It's important to remember that in any analysis of trends, past performance does not guarantee future performance. Just because a customer purchased 200 units of material last year, doesn't mean that they still have the same need this year.
Now that we've reviewed the basic concepts of time series analysis, let's return to the problem faced by the management team: how many robots should be started? We start with a quick examination of the sales data from the last 14 months.
As we observe from the sales data, the robot sales in February had a slight decline from the previous month. Between March and April, a big increase in sales was realized. The percent increase can be measured by subtracting the sales from February from the March, then dividing the result by the sales in February.
Over 79,000 lessons in all major subjects
Get access risk-free for 30 days,
just create an account.
Wow, sales grew over 70% during this month. I'm sure NRAR saw the future as bright moving into April. Using the same formula, the increase in sales for April was 67%. Two months of close to 70% growth secured a new round of investment. Then, May came along. In May, the increase in sales was a meager 10%. Each month following increased by only 10% on average, except for December. In December, sales increased by 20.5%.
What number do we use to forecast robot sales in June? If we examine January and February of the current year, we see a slight reset of the sales number in January. The 10% growth curve began again in February. Should we use 10%? It's probably a good idea.
The increase in sales in December is to be expected and part of a seasonal trend. The fall in January is also part of the seasonal holiday sales. The return to a 10% growth curve in February is encouraging. We should consider both December's increase and January's fall as part of a seasonal trend for planning purposes and exclude the growth rates from our forecast. Drawing a straight line from the first month to the last more accurately represents the trend 10%.
Doing this, we have a growth rate of 10%. Our multiple for forecasting is 1.10. We can either multiply our February sales by 1.10 and repeat that activity three times. Or we can raise 1.10 to the power of three (1.103 = 1.331) and multiply the result times the February sales. Either way, the result is 293.
Our forecast for robot sales in May is 293 units. Production can begin work on the robots now with confidence they will be sold. Of course, we'll need to monitor each month for changes in the growth trend and use a rolling six-month average adjusted for seasonal trends to project upcoming months.
Time series analysis is helpful when projecting future events if the future is expected to be similar to the past. Observation of historical data is likely to result in three different curves: trend, cyclical, and seasonal. A trend is a consistent change in the data in one direction or the other. A cycle will show an increase, followed by a decrease, and a return to an increase. A cycle tends to repeat itself with regularity. Seasonal variations are also cycles, however, they tend to be limited to a specific season of the year, such as the holidays for retailers or the growing season for agriculture.
Did you know… We have over 200 college
courses that prepare you to earn
credit by exam that is accepted by over 1,500 colleges and universities. You can test out of the
first two years of college and save thousands off your degree. Anyone can earn
credit-by-exam regardless of age or education level.