Skip to content
Code to Create GBM model. Top 5% Score
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.
Final Model.R


Code to Create GBM model. Top 5% Score.

alt tag

Kaggle hosts online data science competitions. This competition asks participants to forecast city bikeshare usage in Washington D.C. given weather data. A summary of the competition can be found HERE or by clicking the graphic.

Initially I was inspired by a blog post by Brandon Harris. The main take aways were to consider time of day and weekday as factor variables. These were constucted from datetime variable. This model had a root mean squared logarithmic error of .49523

Efavdb is a cool data science blog. They shared code in python and I converted it to R. Three ideas came from this blog. The first was to use GBM rather than Conditional Inference trees. The second was to tune the hyper-parameters. On future competitions I plan on spending even more time hyper-parameter tuning. The third idea was build seperate models for Regular and Casual users. The python code had a listed RMSLE of .41969

My Final Public Leaderboard score was .37936. (111 out of 3252) I used different hyper-parameters than Efavdb and included more data transforms to improve my score.

You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.