Submissions are evaluated in terms of $$F_\beta \text{ measure, with }\beta=\frac{1}{3}$$between the predicted activities and the real ones of the 6th day between 7am and 19pm.

**N.B.:** the python implementation used for evaluation corresponds to 
```
fbeta_score(solution.activity, prediction_file.activity, 
            average='micro', beta=1/3)
```
This *'micro'* version of the *F* metric is based on the global number of true positives, false negatives and false positives.


## Submission File
For each second of the test set, you must predict an activity from the set of predefined activities: 'A', 'H', 'M', 'P', 'S', 'T, 'kiné' **and 'none'**. The file should contain a header and have the following format:

    time, activity
    '07:00:00','none'
    '07:00:01','none'
    '07:00:02','P'
    '07:00:03','P'
    ...
    '18:59:59','none'
    '19:00:00','none'

## Code File
The code files that you have used to generate your best prediction has to be submitted on campus [here](https://campus2.mines-ales.fr/mod/assign/view.php?id=42081) before Friday the 20th of December (defence day) at 11 pm.

## Report
Short reports (5-6 pages) have to be submitted on campus [here](https://campus2.mines-ales.fr/mod/assign/view.php?id=42085). The reports should summarise your data analysis and your approaches (no code).

# Defence
On Friday the 20th of December the hackathon defences will take place at amphi 064 from 9am to 1pm. All students are asked to attend all defences. 
For each defence, only 1 group member will present the group work during 10min, then all the group members will have to answer the jury questions during 10min. Please focus your presentation on your approach(es), a brief data analysis can be presented as soon as it contains some originality but the subject, context and objectives should not be part of the presentation (we all know the subject).

Here is the defence planning:
9h00	group 4
9h25	group 8
9h50	group 1
10h15	group 2
10h40	group 3
11h20	group 5
11h45	group 6
12h10	group 7
12h35	group 9