Sliding Average Explanation
The term Sliding Average has the same meaning as Rolling Average and Moving Average, or MA for short. I've talked about MAs and MMAs (Multiple Moving Averages) at length over many posts. I like to use MMAs as a reverse crystal ball to find patterns in my web traffic and ad revenue, as well as a means to making an educated guess at future trends.
A moving average is essentially an average value based on a window, usually of time in days, weeks, months, or years. For my blog traffic analysis, I use windows of 28 days (4 exact calendar weeks). The longer the window period, the smoother the trend graph is. However, because I only have less than a year's worth of blog statistics, I'm using 28 days. The window has to be fixed. You can't use 28, 30, 31 days in rotation, depending on the calendar month. You could use a window of 7d, but I feel that this is too small and obscures the real trends. As you've likely read, professional blogging requires a long-term strategy. Hence the reason why I use a 28 d window.
That said, I also use several other windows of 56 d, 84 d, etc., increasing by 28 d each time. These multiple windows give me Multiple Moving Averages (MMAs), and show me both short- and long-term trends simultaneously. If you've ever followed a stock's performance, you may have seen these trend charts. However, they are typically daily charts, with large fluctuations, which obscure long-term performance.
All this may sound scary to those of you that are math-fearing, but it really is pretty easy. For whatever value you are tracking trends for, say daily pageviews, you decide on a window, say 28d, and add up that many consecutive values. So, if your first day of data starts on Oct 1, 2005, your first average is calculated by adding up all your pageviews for Oct 1-28, then dividing by the window size, 28d. This is your first average for your 28d MA. Now repeat for Oct 2-29, etc., until you no longer have 28 days to sum up. Plot the resulting averages over time, against the daily values. Repeat the process for a larger window, say 56 d.
Keep repeating for further larger windows, and you will have an MMA graph. The largest window will show the smoothest curve because the fluctuations have been averaged out. Ideally, your longer-term MA graphs will start to resemble an exponentially increasing curve.
I made promise to post a spreadsheet in which you can plugin in your data, and I haven't forgotten. I'll still do it, hopefully after I've converted this blog over to WordPress. Or Drupal. I haven't decided.
(c) Copyright, 2006-present, Raj Kumar Dash, http://blogspinner.countwordula.com/
Technorati Tags: blogspinner, blogging, pro blogging, multi blogs, web analytics, moving average, sliding average, rolling average, web metrics
A moving average is essentially an average value based on a window, usually of time in days, weeks, months, or years. For my blog traffic analysis, I use windows of 28 days (4 exact calendar weeks). The longer the window period, the smoother the trend graph is. However, because I only have less than a year's worth of blog statistics, I'm using 28 days. The window has to be fixed. You can't use 28, 30, 31 days in rotation, depending on the calendar month. You could use a window of 7d, but I feel that this is too small and obscures the real trends. As you've likely read, professional blogging requires a long-term strategy. Hence the reason why I use a 28 d window.
That said, I also use several other windows of 56 d, 84 d, etc., increasing by 28 d each time. These multiple windows give me Multiple Moving Averages (MMAs), and show me both short- and long-term trends simultaneously. If you've ever followed a stock's performance, you may have seen these trend charts. However, they are typically daily charts, with large fluctuations, which obscure long-term performance.
All this may sound scary to those of you that are math-fearing, but it really is pretty easy. For whatever value you are tracking trends for, say daily pageviews, you decide on a window, say 28d, and add up that many consecutive values. So, if your first day of data starts on Oct 1, 2005, your first average is calculated by adding up all your pageviews for Oct 1-28, then dividing by the window size, 28d. This is your first average for your 28d MA. Now repeat for Oct 2-29, etc., until you no longer have 28 days to sum up. Plot the resulting averages over time, against the daily values. Repeat the process for a larger window, say 56 d.
Keep repeating for further larger windows, and you will have an MMA graph. The largest window will show the smoothest curve because the fluctuations have been averaged out. Ideally, your longer-term MA graphs will start to resemble an exponentially increasing curve.
I made promise to post a spreadsheet in which you can plugin in your data, and I haven't forgotten. I'll still do it, hopefully after I've converted this blog over to WordPress. Or Drupal. I haven't decided.
(c) Copyright, 2006-present, Raj Kumar Dash, http://blogspinner.countwordula.com/
Technorati Tags: blogspinner, blogging, pro blogging, multi blogs, web analytics, moving average, sliding average, rolling average, web metrics
