An Introduction to Trend Analysis

In project management, it is useful to discern trends in the quality data to determine if the project is progressing according to quality expectations. Trend analysis is a technique that tells project managers whether quality goals are being achieved according to the quality management plan.

Trend analysis is a mathematical technique using statistical methods that provide an equation that best fits data in a scatter diagram. Scatter diagrams are simple X and Y axis diagrams with an independent variable, such as time, as the X axis, and the dependent variable as the Y axis. Trend analysis determines the best or most appropriate equation and measures the fit of the equation to the data. Trend analysis is also known as "curve fitting."

Fitting a curve is often done by the least squares method, a mathematical method in which the distance between the data points and a possible line is minimized over its length. This gives the most statistically accurate representation. These lines are often called "regression lines."

Trend analysis is a useful tool for cost and schedule performance, and quality control. The utility of the trend analysis is that it gives a clear and understandable indication of change caused by every incremental change of the independent variable. One of the more useful functions of trend analysis is predicting, or forecasting.

The different lines mean a variety of different things could have occurred in a process. Line and curve shapes indicate whether a process is behaving according to the quality control norms.

• Lines of positive correlation - Lines of positive correlation indicate the desired value y is increasing. This is good if improvement is sought, but bad if the line continues past a specified value.
• Lines of negative correlation - Negative correlation indicates y is decreasing. This is good if the tolerance of a process is coming closer to a desired value, but bad if that same value is exceeded.
• No correlation - A diagram with no correlation means the data is inconsistent. The process is out of control, and immediate steps are necessary to bring the process under control.
• No slope lines - A line with no slope means there is no change. This is indicative of a stable process.
• Curvilinear line - Curvilinear lines indicate a cyclical process or a process decreasing or increasing at a non-uniform rate. Cyclical patterns indicate a possible worn out process. A curvilinear line indicates a complex relationship with the independent variable.

Of all the different lines, curved lines are the most difficult to make conclusions from because of their shape. Other statistical analysis must be used to determine at which point the objective value has been met or exceeded.

Trend analysis allows project managers and teams to predict a pattern and come up with a formula that accurately reflects a data set. As long as the appropriate quantity of data have been selected, accurate predictions can be made of a process. Trend analysis is also useful for determining at which point a quality concern may become an issue based on historic data. Trend analysis is often useful when used in conjunction with other tools and techniques.