Run charts are a great technique to track data over a period of time. This graphic includes data points throughout a time series. With this data we can identify trends and look for shifts in the data over time. This is a great way to determine whether or not the process is under control, or perhaps a warning if the process is drifting with respect to the desired results. Run charts can also be very useful to validate that a corrective action we’ve taken to implement Lean or Six Sigma improvements is actually working. After implementing a change, we can reference the run chart to observe whether the desired effect is occurring. Is the data improving over time as the process continues over time? A run chart is a wonderful way to monitor progress and process improvement, ensuring that a process is staying in control. The value of the run chart extends to observing results and maintaining control while we’re making improvements.
As we create a run chart, there are different elements we will include. First we have the left axis with a numeric score scale of quantities or percentages, or any other attribute you may be measuring. Next we have the median line. This would be the average of the high and the low values for all the data in our data set. Across the bottom, we have the time series. This could be hours, days, or any other increment of duration. Then we have the data points for the outputs, allowing us to measure an attribute for each unit that was created at any given time in the process. An attribute could be any value such as weight, length, battery life, etc., for the unit that was just created. If you think of a production line, these could the units coming off a conveyer belt one at a time. As they are produced, we take note of the time and measure of the attribute of the output to determine where it falls on the scale. Then we plot that data point and move on to measure the next unit.
There are some basic decision rules that we should pay attention to when using run charts. In general we should expect the data points to be relatively stable, alternating on either side of the median line, and always remaining within a reasonable distance to the median. If you notice a trend where seven points or more data points in a row fall above or below the median, it’s an indication that something has changed in the process. In the graphic on the left side, there’s an event where a series of seven or more data points in a row has fallen above the median. This should be investigated.
The run chart on the right shows an upward trend over time. If we examine the data over a series of months, there is a trend where the data is gradually moving upward.