With the trained time series model of Capital Bikeshare capacity predictions, one can predict the capacity of a given station with some help visualization of current and historic station capacity. The notebooks are avaiable to view here.
The historic station status data was manually downloaded from capital bikeshare's data bucket. From there it was read into and concatenated into one pandas dataframe(343.4+ MB). For the snapshot of bike station statuses I used urllib.request to download and parse the xml into a pandas dataframe. From their the two dataframes were seperatly transformed before combining. Lastly, outage data was downloaded from the Capital bikeshare outage webpage to later refine the combined dataframe for a time series analysis. Finally, in the mapbox-jupyter notebook I read in the processed data from the analysis notebook and changed the data to geojson for mapping.
- In the Bikeshare Status Analysis notebook I have standard visualizations of station capacity preceeding the various time series plots.
- Percent capcity histogram
- Percent capacity Boxplots
- While in the mapbox-jupyter notebook I have maps of Graduated Circle map of the current bike number avaiable at a station noted by a dot thats radius reflect the total capacity of that station. Additionally, I have a heat map that reflects the average duration of bike trips starting at that station by increasing radius.
- os, pandas, numpy,
- urllib, shutil, xml,
- matplotlib, dateutil, statsmodels,
- datetime, sklearn, warnings
- Matplotlib, Seaborn, Mapbox