Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
exercises:start [2014/08/14 15:25]
hverbeek [Exercise 1]
exercises:start [2014/08/14 15:31] (current)
hverbeek
Line 4: Line 4:
  
 {{page>:​exercises:​introduction&​firstseconly&​noeditbtn}} {{page>:​exercises:​introduction&​firstseconly&​noeditbtn}}
- 
- 
 {{page>:​exercises:​exercise1&​firstseconly&​noeditbtn}} {{page>:​exercises:​exercise1&​firstseconly&​noeditbtn}}
-===== Exercise 2 ===== +{{page>:​exercises:​exercise2&​firstseconly&noeditbtn}} 
- +{{page>:​exercises:​exercise3&​firstseconly&noeditbtn}} 
-Repeat the questions of [[#​exercise_1|exercise 1]] on the traces shown below which are stored in the file exercise2.xes. +{{page>:exercises:exercise4&firstseconly&noeditbtn}} 
-  1x Case1 A C D +{{page>:​exercises:​exercise5&​firstseconly&noeditbtn}} 
-  1x Case2 B C E +{{page>:exercises:exercise6&firstseconly&noeditbtn}}
-It might be nice to play around with the settings of the Transition System Miner and see what the impact is on the resulting transition system. +
- +
-<hidden Solutions (click to open or close)>​ +
-  -- \\ +
-  -- Your results should be similar to the screenshots shown in the figures below.{{ :exercises:prom-6.11.jpg?​direct&​570 |}}{{ :​exercises:​prom-6.12.jpg?​direct&​570 |}}{{ :​exercises:​prom-6.13.jpg?​direct&​570 |}}{{ :​exercises:​prom-6.14.jpg?​direct&​570 |}} +
-  -- If you open exercise2.xes in a text editor with XML syntax highlighting you get something like in the figure below.{{ :​exercises:​prom-6.15.jpg?​direct&570 |}} +
-  -- The expected result is shown in the figure below.{{ :​exercises:​prom-6.16.jpg?​direct&570 |}} +
-  -- The expected result is shown in the figure below.{{ :exercises:prom-6.17.jpg?​direct&​570 |}} +
-</​hidden>​ +
-===== Exercise 3 (extra) ===== +
- +
-You might want to repeat the questions of [[#​exercise_1|exercise 1]] for exercise3.xes, of which the content is shown below. The result in a very nice transition system and petri net. No answers are provided for this exercise however. +
-  1x Case1 A C E G +
-  1x Case2 A E C G +
-  1x Case3 B D F G +
-  1x Case4 B F D G +
- +
-Solutions not provided since repeating the solution approaches of exercises 1 and 2 would provide you with the correct answers. ​  +
-===== Exercise 4 ===== +
- +
-For this exercise we will use the traces shown below which are stored in exercise4.xes. +
-  1x Case1 a b c d f +
-  1x Case2 a c b d f +
-  1x Case3 a b d c f +
-  1x Case4 a c d b f +
-  1x Case5 a d e f +
-  1x Case6 a e d f +
- +
-  -- Since it is always a good idea to inspect the event log at hand, inspect exercise4.xes to get an idea of its contents. What is remarkable about this event log? +
-  -- Try to construct a Petri net by hand which can replay the traces recorded in exercise4.xes. Use 6 transitions that represent the actions A through F. Try to allow as little extra behavior as possible (e.g. no '​flower'​-nets). +
-  -- Now create three Petri nets using the following plug-ins: +
-    - As in exercises [[#​exercise_1|1]]-[[#​exercise_3|3]],​ first create a transition system from exercise4.xes and then create a petri net from the transition system; +
-    - Run the alpha-algorithm on exercise4.xes;​ +
-    - Run the ILP Miner on exercise4.xes. +
-  -- Compare the three Petri nets created by the different algorithms. Are there any fundamental differences or are they bisimilar? Why do you think that there is a difference between (some of) the Petri nets? +
-  -- Now run the Heuristics Miner on exercise4.xes. Is this a Petri net? What is different?​ +
-    - Create a new visualization of the HeuristicsNet that shows the join and split semantics. Is this a similar net as the Petri nets from the previous steps? +
-  -- Now run the Genetic Miner on exercise4.xes. Why is the result with the default settings so poor? Try to improve the setting and get a result with a fitness of at least 0.8. +
- +
-<hidden Solution (click to open or close)>​ +
-  -- One of the most remarkable things is that all events have the same date and time. +
-  -- See the results of the Petri net discovery plug-ins performed for the next question in the figures below.{{ :​exercises:​prom-6.20.jpg?​direct&570 |}}{{ :​exercises:​prom-6.21.jpg?​direct&570 |}} +
-  -- \\ +
-    - The expected results are shown in the figures below.{{ :exercises:prom-6.18.jpg?​direct&540 |}}{{ :​exercises:​prom-6.19.jpg?​direct&540 |}} +
-    - The expected result is shown in the figure below.{{ :​exercises:​prom-6.20.jpg?​direct&​540 |}} +
-    - The expected result is shown in the figure below.{{ :​exercises:​prom-6.21.jpg?​direct&​540 |}} +
-  -- The resulting Petri nets of the alpha-algorithm and the ILP Miner are the same. The one converted from the Transition System Miner result is different however. The reason for this is that the alpha-algorithm and the ILP Miner only allow for each event to be present only once. In the transition system events are present multiple times. During the conversion of the transition system to a Petri net not all events are merged, therefore the resulting Petri net looks different but describes exactly the same behavior. +
-  -- The result of the Heuristics Miner, as shown in the figure below, is not a Petri net. The net only indicates direct relationships without specifying the split and join semantics.{{ :​exercises:​prom-6.22.jpg?​direct&​570 |}} +
-    - To visualize the Heuristics net from the previous action with the split and join semantics we need to select the visualization option as shown in the tope figure below. The resulting net then shows the split and join semantics, as shown in the bottom figure below. The resulting net is essentially the same as the Petri nets that resulted from the alpha-algorithm and the ILP Miner.{{ :​exercises:​prom-6.23.jpg?​direct&​540 |}}{{ :​exercises:​prom-6.24.jpg?​direct&​540 |}} +
-  -- In the top figure below the result with the default settings for the Genetic Miner is shown. We increased the number of iterations from 2 to 200 to get a model with a fitness of 0.875. In the bottom figure below the resulting model is shown.{{ :​exercises:​prom-6.25.jpg?​direct&​570 |}}{{ :​exercises:​prom-6.26.jpg?​direct&​570 |}}  +
-</hidden> +
-===== Exercise 5 ===== +
- +
-For this exercise we will use exercise5.xes which is the result of a simulation. +
-  -- Start with a log inspection on exercise5.xes +
-  -- Create some proces model(s) using two (or more) of the plug-ins from the previous exercises. +
-  -- Run the dotted chart analysis plug-in on exercise5.xes and try to answer the following questions:​ +
-    - What do you see using the default settings? What does each row, column and dot represent?​ +
-    - What does the diagonal line from the top left to the middle of the last row represent?​ +
-    - Using the default settings, what do you think is the meaning of the row of dots that is located below and to the left of the diagonal line? +
-    - Can we see a clear division of tasks that are executed in the beginning and (repeated) near the end of the process? +
-    - Change the settings to "Time option: Relative(Time)"​ and "Sort By: Actual duration"​. What do we see now? +
-    - Change the settings to "​Component type: Originator"​ , "Time option: Actual"​ and "Sort By: None". What do we see now? +
-    - Using these settings, can we distinguish groups of users doing similar tasks? +
- +
-<hidden Solutions (click to open or close)>​ +
-  -- There is not really anything special to discover during log inspection in this case. +
-  -- The result of the alpha-algorithm is shown in the top figure below and the result of the Fuzzy Miner in the bottom figure below. Each plug-in should come up with similar process models.{{ ​:exercises:prom-6.27.jpg?​direct&​570 |}}{{ :​exercises:​prom-6.28.jpg?​direct&​570 |}} +
-  -- \\ +
-    - When you run the Dotted Chart on exercise5.xes you get a result similar to the figure below, although your colors can be different. Each horizontal line is one case, as the "​Component Type" is set to "​Instance ID". On the x-axis on top you see dates so from left to right you progress through time. Each dot is an event as recorded in the event log. Each color refers to a particular activity, as the "Color By" option is set to "Task ID".{{ :​exercises:​prom-6.29.jpg?​direct&540 |}} +
-    - The diagonal line highlighted in the figure below indicates the start of new cases. Since it is logical that cases with a higher identification number start later in time, the line of start events will not be vertical but actually progress diagonal.{{ :​exercises:​prom-6.30.jpg?​direct&540 |}} +
-    - In the figure above these sections of dots are circled. If you select dots in ProM you will see that the identification number of the case is between 2 and 9. The sorting algorithm sorts the numbers on their first digit so the order in the graph is 19-2-20-21. Can you spot case number 100? +
-    - Yes, if we look again at two figures back we see that at the beginning of cases '​red'​ events are executed which do not appear later in time for a case. +
-    - The resulting dotted chart is shown in the figure below. Time is made relative from the first event for each case and cases are sorted with the shortest on top. This gives a good indication of the time distribution amongst cases. We can for instance see that some cases are processed within 30 or 60 days. The longest case however lasts for 720 days after its first registered event.{{ :exercises:prom-6.31.jpg?​direct&540 |}} +
-    - The result is shown in the figure below. What we see now is which user executes which activity at what time.{{ :​exercises:​prom-6.32.jpg?​direct&540 |}} +
-    - It appears that there are 4 groups of users doing similar tasks. Anne and Mike do similar tasks as does the group of Carol, John, Mary, Pam, Pete, Sam and Sara. Then there is Wil which only performs one type of tasks and he is the only one that performs this task. Then there is the '​_INVALID'​ '​user'​ which `executes'​ all kinds of activities. +
-</​hidden>​ +
-===== Exercise 6 ===== +
- +
-For this exercise we will use exercise6.xes which is the result of a simulation.+
  
-  -- Start with a log inspection on exercise6.xes. What is remarkable about this event log? 
-  -- Optional: run the Dotted Chart Analysis on this event log to visualize it in another way. 
-  -- Create some process model(s) using two (or more) of the plug-ins from the previous exercises. 
-  -- Run the Fuzzy Miner on the event log, using the default settings. Now try to answer the following questions: 
-    - What is remarkable about the resulting process model compared to the models discovered by the other plug-ins? 
-    - The Fuzzy Miner tries to aggregate low frequent activities into clusters and only show edges which are significant enough. Using this brief explanation,​ can you indicate why the resulting process model looks so different? 
-    - Read the [[https://​svn.win.tue.nl/​trac/​prom/​browser/​Documentation/​Package%20Fuzzy.pdf|documentation on the Fuzzy Miner]]. ​ 
-    - Play around with the settings of the Fuzzy Miner (the sliders you see on the right). What do you need to change to get closer to the models as discovered by the other plug-ins? 
-    - Why do you think this model is still not exactly the same as the original model? 
-    - On what kind of logs would the Fuzzy Miner be useful? 
-    - Can you now think of some benefits but also some weak points of the Fuzzy Miner? 
-  -- Run the Transition System Miner on the event log. Be sure to set the "​backward keys" to both "Event Name" and "​Lifecycle transition"​ and not "​Resource",​ as is shown in the figure below.{{ :​exercises:​prom-6.jpg?​direct&​570 |}} 
-  -- From the resulting transition system, press the "​Play"​ button on the top right. Select the "​Transition system analyzer"​ plug-in. Now click the "​Log"​ object in the "​input"​column and select exercise6.xes. Now run the plug-in (press "​Start"​) and try to answer the following questions: 
-    - How long does a case take on average? And what are the minimal and maximal durations recorded? On how many ways can you get this information from this plug-in? 
-    - What do the red states indicate? And what do the red arrows indicate? 
-    - On the top left, change the "Color By:" setting from "​Sojourn"​ to "​Elapsed"​. Why is part of the process colored red and why are the last two events yellow? 
-    - If we change the setting to remaining we see the first events colored red and 1 event colored yellow. Can you explain this? And why do you think that the outgoing arrow of the yellow event is colored red? 
  
-<hidden Solutions (click to open or close)> 
-  -- The most remarkable thing is that the event log includes events that are executed between 2006 and 2018. It also contains events of 2 types: start and complete. This allows for analysis of waiting and execution times. 
-  -- You should know how to do this but we won't discuss things that you might discover. 
-  -- The Heuristics Miner and alpha-algorithm come up with correct models. If the Transition System Miner is run with the settings as explained later on, the model is also correct. When run with the default settings you end up with 2 end events which in essence are the same activities. 
-  .. The ILP Miner comes up with the result shown in the figure below which is slightly different. On the one hand it is simpler to read, on the other hand it allows for more behavior than recorded in the event log.{{ :​exercises:​prom-6.33.jpg?​direct&​570 |}} 
-  -- Running the Fuzzy Miner with the default settings will result in the model as shown in the figure below.{{ :​exercises:​prom-6.34.jpg?​direct&​570 |}} 
-    - The resulting process model is entirely sequential while the other process models discovered from the same log are not. 
-    - The Fuzzy Miner removes edges which are considered to be not relevant enough to show. 
-    - \\ 
-    - The most important setting you need to change is the "Edge cutoff"​ as shown in the figure below. Changing other settings might also improve the result.{{ :​exercises:​prom-6.35.jpg?​direct&​570 |}} 
-    - The Fuzzy Miner takes many metrics into account when deriving nodes and edges, most of these are not used by other plug-ins. All metrics used by the Fuzzy Miner can be seen and changed by selecting the Metrics tab when starting the Fuzzy Miner. 
-    - The Fuzzy Miner is particularly useful on large, unstructured and/or noisy logs. 
-    - The main benefit is that it can cope with unstructured event logs. The main weak point is that the model discovered by the plug-in does not really contain any semantics and only indicates '​relationships'​ between nodes/​events. 
-  -- \\ 
-  -- \\ 
-    - You can answer this question if you select the first state and then inspect the metrics, as shown in the figure below. The remaining time after the first state is 1118 days and 4 hours on average. The minimal remaining time is 8 days while the longest remaining time is 4462 days. Another option is selecting the last state of the transition system and looking at the elapsed row in the metrics table.{{ :​exercises:​prom-6.36.jpg?​direct&​570 |}} 
-    - States and arcs are colored blue, yellow or red depending on the relative time spend there. In the figure below state 4 is red and selected. On the bottom of the screen some metrics are shown. Here we can see that the sojourn time is very high, hence the red color of the state. The red arrow is also colored red for the same reason. Apparently the time between starting and completing the registration of a claim takes 510 days on average. State 8, assessment of a claim, also takes a long time, 599 days on average.{{ :​exercises:​prom-6.37.jpg?​direct&​570 |}} 
-    - In the figure below this option is chosen and the second-last state is selected. The last two events are colored yellow because there are different ways to get there. A case is for instance already ended when insuffcient information is available or when the likelihood of the claim is deemed to low/high (we don't know that). The part that is colored red is the path that a handled claim takes which takes longer than to be rejected.{{ :​exercises:​prom-6.38.jpg?​direct&​570 |}} 
-    - The figure below shows the result of this option with the one yellow state, state 8, selected. Of course, now the first part of the process is red because it can take a long time before the process is completed. State 8 is colored yellow because activity Assess Claim can take quite a while but does not qualify for the color red yet.{{ :​exercises:​prom-6.39.jpg?​direct&​570 |}} 
-</​hidden>​ 
  • Show page