JY Tigli Homepages

Cette page est en lecture seule. Vous pouvez afficher le texte source, mais ne pourrez pas le modifier. Contactez votre administrateur si vous pensez qu'il s'agit d'une erreur.
====== MAIN INFORMATIONS about the Project ======

====== Page du groupe 3 ======

===== Members ======
  * Benjamin Benni SI5, AL (bb102748)
  * Hugo Simond SI5, IHM (sh003984)
  * Vincent Montalieu SI5, GMD (mv305347)

===== Project =====

===== Title of the Project =====
How much time do I work ?

==== Main idea ====
The main idea of this project is to show how much time a user spend at home, at work or the stadium.

==== Data collect  ====
The user install our app on his smartphone (that is a good start). He starts the app for the first time (yes, he must). The app is asking him to choose how many place he visits frequently (minimum 2, maximum 10). Let's call this variable "C".

He click on "OK" and the app starts to track his GPS location.
Every time in moves from "M" meters, the system throws an event, saying that the user has changed his location. We can say that "Mister X was at P position at T".


==== Data analysis ====

=== Building clusters ====
- We start by building clusters on the set of his GPS locations. We establish "C" clusters.
  
  - Now that the clusters are built, we must put a label on it. To labelize it, we evaluate following conditions :
  - If a cluser has a location retrieved at midnight and a location retrieved at 03:00am so let's call it "HOME"
  - If a cluser has a location retrieved at 9am and 11am OR 2pm and 4pm, so let's call it "WORK"
  - Other clusters are nammed "Ci" where i can take values from 1 to C-2

Next, the user have to indicate "Ci" <==> "A_SYMBOLIC_NAME".


=== Compute duration spent on every cluster ===

Now that every cluster are built, we analyze the clusters. We find the most ancien and the most recent location in the cluster. We deducts the time spend on each place.

The user can see that he spent "10h at home", "7h at work" and "2h at the gym" if we assume that he indicates C = 3 clusters at the first step.

===== Developped SOFTWARE =====

==== Screenshots and GUI description ====

<TODO>
==== SOFTWARE PACKAGES of the Project ====

<TODO>
==== Required SOFTWARE Environment ====

You need to have previously installed Visual Studio 2013 (Community edition will do the trick). The zipped solution will be directly importable from VS. You will also need Windows Phone 8.1 emulator to test our app.
We tested it on Windows 10 but it should work on Windows 8.1 and 8.0 platforms.

To build it we used Windows Phone 8.1 Silverlight project so you need to have WP8.1 SDK installed if you want to change anything in the source code.
==== How To build solution from project sources ====

Import the solution sources into Visual Studio (File -> Open -> Open File -> open the solution file).
From now, you just have to click on "build solution" and launch it. If you selected to deploy the app on an emulator, it will open the emulator for the specified device.
===== HOW TO USE IT and RESULTS =====


==== How to use it ====

The first time you start the app, it will ask you if you allow tracking your GPS position.
Please say yes. If you answer "no", the app is just USELESS.

Next screen displays the main screen of the app : 
on the top right corner, you have a map showing your current location
on the top left corner it shows you your current latitude and longitude and cluster related
on the bottom on the screen, it displays the different locations (Home / Work / Sport) and time spent on each location (i.e. on each cluster).

==== Results =====

  * Clustering works. We define 3 coherent clusters.
  * The time spent on each cluster works.
  * **But label put on each cluster is completly wrong.** 


Since the algorithm used to build clusters can change indexes for clusters built, cluster 0 will be "HOME" and next time cluster 0 will be "WORK". 

So it has a random behavior. To fix it, we should do the following :
build 3 clusters from collected GPS locations (done already), label each location on each cluster with "HOME" or "WORK" or "SPORT". 

Next, we should count how many "HOME" location are tagged in cluster 0, cluster 1 and cluster 2. The cluster that has the most labels "HOME" should be declared as "Cluster HOME". The same logic should be applied to "WORK" and "SPORT". We try to do it but not achieved it (17/11/15).


==== Video ====
==== Tutorial ====