project.Rmd

---
title: "Project"
author: "Jonathan Rosenblatt"
date: "March 24, 2015"
output: html_document
---

Here are the guidelines for the course's concluding project.
The project is based on the [Bike Sharing Demand](https://www.kaggle.com/c/bike-sharing-demand) competition at Kaggle.
You are required to submit a prediction to Kaggle, and a report on the process to me.

# Dates
Teaming up: no later than __27.4.2015__.
Submit prediction to Kaggle: __29.5.2015__.
Submit report to Jonathan : __26.6.2015__.

Recommended time-table:

1. During the Passover vacation, download the data. Make sure you can load it and practice `dplyr` and `lubridate` on it.
2. After Passover, find your team and notify me.
3. Keep revisiting the data as we progress and study new techniques. Don't leave everything to submission date.


# Guidelines

1. Your task is to participate in the [Bike Sharing Demand](https://www.kaggle.com/c/bike-sharing-demand) competiton. The competion ends on __29.5.2015__ when you will have to submit your predictions to Kaggle.
2. You can do so in pairs, or trios.
3. By the end of the course you will need to submit to me a report documenting the process.
  - No longer than 8 pages (not including appendices).
  - Submitted by mail which includes:
    - A PDF file with the report. 
    - Author names and IDs.
  - Should contain the sections:
    - Background: Some background on the competition. 
    - Scoring: The scoring criterion in the competition. What loss function with what data?
    - The data: What data was provided for learning? What files in what formats? Which variables? How did you handle them?
    - Algorithms: Which learning algorithms did you try?
    - Results: What score did you achieve? What was your ranking in the competition?
    - Discussion: Why were you successful/unsuccessful? What other ideas would you have liked to try? What were the major challenges? 
  - Code should be added in appendices. 
    4. Feel free to use the courses forums for questions. Especially regarding the use of Kaggle and R. Make sure however, that you do not share your solutions. 
    5. Any non trivial choices you made in the project need to be justified: tell me "why", not only "what".