Let's start solve problems challenges on Kaggle

This is presentation/training prepared for DataKRK. I'll be happy if it help you start solve challenges on Kaggle 😄!

How to start use it?

1. git clone https://github.com/slon1024/lets-start-solve-problem-on-kaggle
2. cd lets-start-solve-problem-on-kaggle
3. git fetch

Run ipython (jupyter) notebook (inside directory lets-start-solve-problem-on-kaggle): ipython notebook.
Note: you need install ipython before (and some other needed packages) or just install anaconda.

After this you will have all three steps (in different branches, initial step is on the master).

1. Step 1 is on the branch step1 (git checkout step1)
2. Step 2 is on the branch step2 (git checkout step2)
3. Step 3 is on the branch step3 (git checkout step3)

Note: solution is inside this file bike-sharing-demand.ipynb. After switched a branch you should refresh your notebook in a browser.

Where is train.csv file?

You should download this from Kaggle site (Bike Sharing Demand - data).

Quick review scripts

Note: github allowing you do a qucik review (of course readonly). For bigger scripts this works very slowwwwly.

1. Step 0 (initial step)
2. Step 1 (improve evaluation a model)
3. Step 2 (understand better data and improve a model)
4. Step 3 (feature engineering and tuning model)

I recommend to you follow step by step.
If you have any questions or ideas, please let me know.