ISSS608 2016-17 T1 Assign3 Goh Kar Hock Raymond
Contents
- 1 Data Preparation
- 2 Analysis Methods
- 3 Analysing of 10 Patterns
- 4 Cluster 1 from Friday
- 5 Cluster 2 from Friday
- 6 Cluster 3 from Friday
- 7 Cluster 4 from Saturday
- 8 Cluster 5 from Saturday
- 9 Cluster 6 from Saturday
- 10 Cluster 7 from Saturday
- 11 Cluster 8 from Sunday
- 12 Cluster 9 from Sunday
- 13 Cluster 10 from Sunday
- 14 Vandalism Detection
Data Preparation
JMP is the main programme used to prepare and join the data into the required file formats for reading in Gephi and Tableau.
The 3 communications data files were needed to be prepared before running in Gephi. The data was summarized based on the grouping of three variables (from, to and locations). The headers for the column were then changed to the headers required for Gephi to read, such as “from” -> “Source” and “to” -> “Target”. A new column “Type” with the data set as “directed” was created. After which, save the file as a .csv file for input into Gephi to study the communication network.
Once the clusters of nodes were identified and extracted in Gephi, it will be saved in .csv file format. Then using JMP, to extract the necessary movement data for the identified nodes for further analysis,
Analysis Methods
JMP was the main programme used to prepare and join data for analysis. In addition, it was also used to analyse the communication pattern of a particular trend identified in Gephi.
Gephi was used to analysis the clusters of communication patterns over the 3 days. The programme was also used to isolate and identify clusters with similar communication trends internally and externally.
Tableau was used to analysis the movement data of the nodes identified. The movement pattern could be matched against the park map and have a good understanding of where the customers had been to over time.
Analysing of 10 Patterns
There were 3 common nodes that majority of the park customers would communicate with within the park.
Node 1278894 (medium brown size circle) – park’s main broadcast of messages when the customers first arrived at the park.
Node 839736 (medium red size circle) – park’s main communicator with the park’s customer for the game services throughout their stay in the park.
Node “External” (big yellow size circle) – Main node for the customers to communicate with people outside the park.
Cluster 1 from Friday
Cluster 1 communicate with the park’s game services constantly and also internally within the group.
https://public.tableau.com/profile/publish/Cluster1_0/Dashboard1#!/publish-confirm
From the movement data seen in tableau, this cluster tends to stick together as a group and visited the Kiddie Land very often. Hence this cluster should be a group of young people who were friends.
Cluster 2 from Friday
At level 1 communication, Cluster 2 was seen to be communicating among the group which was widely spread out and not communicating to the park’s services. There were a few groups of people visited the park as a group, and were loosely linked some other individuals who also visited the park but were not part of the group.
https://public.tableau.com/profile/publish/Cluster2_0/Dashboard1#!/publish-confirm
From the movement data seen in tableau, there were several different groups moving in different directions. Hence I conclude that this cluster was made up of a collective of different groups who knew each other but did not plan to visit the park together.
Cluster 3 from Friday
Cluster 3 was a small group of people who visited the park together. However from the movement data, it seem that there were 2 different age group of people who preferred to play different attractions.
https://public.tableau.com/profile/publish/Cluster3/Dashboard1#!/publish-confirm
Cluster 4 from Saturday
Cluster 4 is a group of people moving around together. Of interest to note was there was a high level of communication within the group at the Wet Land and also communicate outwards to people outside the park while they were in the Wet Land.
https://public.tableau.com/profile/publish/Cluster4/Dashboard1#!/publish-confirm
Cluster 5 from Saturday
Cluster 5 had demonstrated similar traits as group 4. Similarly there was a high level of communication within the group at the Wet Land and also communicate outwards to people outside the park while they were in the Wet Land.
https://public.tableau.com/profile/publish/Cluster5/Dashboard1#!/publish-confirm
Cluster 6 from Saturday
Cluster 6’s communication was not as tightly knitted as Cluster 4 and 5 but the people were seen to be moving around the park as a group. Though there was a high level of communication within the group at the Wet Land, but this cluster had communicated lesser outwards to people outside the park while they were in the Wet Land.
https://public.tableau.com/profile/publish/Cluster6/Dashboard1#!/publish-confirm
Cluster 7 from Saturday
Cluster 7 had shown a high level of communications within the group at the Wet Land, and communicated significantly outwards to people outside the park while they were in the Wet Land. This group tends to move together in the park playing things in general.
https://public.tableau.com/profile/publish/Cluster7_0/Dashboard1#!/publish-confirm
Cluster 8 from Sunday
Cluster 8 was a small group of 3 persons who only communicate with external at level 1. Interestingly, this group of people don’t communicate with the park’s games services at the first level while visiting the attractions. Point to note was this group of people has lesser movement as compared to the previous groups.
https://public.tableau.com/profile/publish/Cluster8/Dashboard1#!/publish-confirm
Cluster 9 from Sunday
Cluster 9 is similar to Cluster 3 which have 2 different group of people who preferred to go for different attractions. It was noted that communications within the cluster was lacking as compared to the previous few groups.
https://public.tableau.com/profile/publish/Cluster9/Dashboard1#!/publish-confirm
Cluster 10 from Sunday
Cluster 10 had a high concentration of communication pattern but this cluster did not move around the park together. Similarity this cluster had shown a high level of communications within the cluster at the Wet Land, and communicated significantly outwards to people outside the park while they were in the Wet Land.
https://public.tableau.com/profile/publish/Cluster10_0/Dashboard1#!/publish-confirm
Vandalism Detection
A surge in in-degree to External nodes on Saturday and Sunday from the Wet Land (Purple Lines), as compared to the in-degree of External nodes on Friday. Hence it could be concluded that the vandalism was detected on Saturday in the Wet Land.