Table of Contents

Basic Performance Analysis

The basic performance analysis calculates peformance measures such as execution (working) time, waiting time, etc. and displays the values with several kinds of graphs.

Preparation

  1. Open an event log file in ProM (The log should have time stamps)
  2. Choose Basic Performance Analysis from the Analysis menu

Settings

The Basic Performance analysis plug-in consists of two parts. The first part is the configuration of performance measures and the second part is displaying result with several graphs. Figure 1 shows the configuration panel.

Basic Performance configuration

Figure 1. The configuration panel of the Basic Performance analysis plug-in

The first three checkboxes allows selecting dimensions for the performance measures for tasks (i.e. execution time, waiting time). User can select task, originator, or instance dimension. For example, if user selects the originator dimension, the execution times can be displayed along with originators. If more than one dimension is selected, the result can be displayed considering two dimensions.

Timezone Offset is the timezone offset of the place where a process log is gathered. You can find this information at a timestamp in the log. The time stamp in a log is influenced by the location. For example, if a log is generated in Eindhoven, the time stamp has two hours of offset value. This raises a problem, when you analyze the log in a different time zone. Note that Timezone offset is used when you take into account working hour. Thus if you do not consider working hour, you can ignore this value.

Remove outliers enables users to remove outliers after calculating performance measures. User can provide lower bound and upper bound.

Considering working hour takes into account office hour. For example, assume that a task is started 16 hr and is ended 10 hr on the next day. If we consider working hour, the execution time is 2 hrs. Otherwise, it is 18 hrs (i.e. 24 -16 + 10).

User can consider weekends/holidays. For this option, if there is no event in a day, we assume that this day is either a weekend or a holiday.

After choosing proper options, the start calculation button should be pressed to get a result.

Results

The result panel consists of several options and resulting graph. There are two major performance measures. They are measures for tasks (processing) and measures for instance (throughput). And it provides 9 kinds of graphs such as Bar chart, Pie chart, Spider chart, Box-and-Whisker chart, Gantt chart, Time chart, Day-Hour chart, Meter chart, and text view of the result.

Bar Chart shows the result with a bar graph.

barchart

Figure 2. Bar chart

Pie Chart shows the result with a pie graph.

piechart

Figure 3. Pie chart

Spider Chart shows the result with a spider graph.

Spider Chart

Figure 4. Spider chart

Box-and-Whisker Chart shows the result with a Box-and-Whisker graph. A box-and-Whisker shows a minimum, a maximum, a median, an average, a 1st-quartile, and a 3rd-quartile values.

Box-and-Whisker Chart

Figure 5. Box-and-Whisker chart

Gantt Chart shows the result with a Gantt chart.

Gantt Chart

Figure 6. Gantt chart

Time Chart shows the result with a Time graph. In a time graph, x-axis refers to time from the start of the log to the end of it. y-axis shows the number of working events or aggregated working hour along x-axis. For example, in the Figure 7, thw aggregated working hour of the first day (from the first 0:00 to the second 0:00) is about 4.50 and that of the second day is about 4.00. It means that on the first day, 4.50 hours of working time are observed and the second day has 4.00 hours of working hours in the log.

Time Chart

Figure 7. Time chart

Day-Hour Chart shows the density of works on a day-hour matrix.

In a Day-Hour chart, x-axis refers to a day from 0:00 to 24:00. Y-axis shows dates from the start to the end of a process log. A box on the chart represents a time unit and the color of it shows the density of works.

Day-Hour Chart

Figure 8. Day-Hour chart

Meter Chart shows the result with a Meter graph. A meter graph consists of four ranges. The white range shows the area where there is no relevant values.

The green, yellow, red represent normal, warning, critical zone respectively.

Meter Chart

Figure 9. Meter chart

Text View shows the result with a table.

Text view

Figure 10. Text view