gumdropsteve/turbo-telegram

GPU accelerated Jupyter dashboard.

turbo-telegram

turbo-telegram's Issues

if/elif/else of map outputs is too complicated

issue

• code is difficult to follow
• checking that focused dataframe has enough rows to display properly should not be necessary with increased table size

fix

• clean up & limit if/elif/else statements of taxi map outputs

cuxfilter?

How can we integrate cuxfilter?

2015 Data Style (what's the effective range?)

As seen with 3e2f3b3, the data for the first half of 2016 can be downloaded & preprocessed just like that of 2015. Is there any other data in the effective range? I.e. is pre-2015 data recorded the same?

If so, let's add it.

[BUG] n_riders & tol_cost input dashboard error

issue

• adjusting or inputing values into the n_riders or tol_cost boxes of the input dashboard submits an invalid query results in error
• both input and output dashboards continue to work, but error displays below input dashboard as part of the cell's output

expected behavior

• do not adjust output dashboard until query is valid
• or give user time to input before processing

How does this fit into e2e workload?

Can we produce a full end-to-end demo using the NYC Taxi dashboard for EDA?

Maybe something like:

• download & preprocess w/ download_data.ipynb
• explore data with taxi_dashboard.ipynb
• do some parameter tuning w/ ??
• hand off to model w/ ??

Clean up data_download.ipynb

What's everything after the "TOTAL TIME" output doing there?

SQL based

Minimize use of cuDF methods & other adjustments to BSQL results

Change graph columns

• I want to be able to substitute in different columns for the 2 scatter plots
• This may help with starting a sort of API conversion

[BUG] Coordinates not converted (Feb & Mar 2015)

Issue

• coordinates for data for Feb & Mar of 2015 have not been converted for HoloViews plots
• pickup/dropoff longitude/latitude columns not renamed to x/y format

Resolve with

• datasets were added w/ #3
• here's the file where conversion can be found

More data please

I want to use NYC dashboard with more data but I'm unable to get it to display properly

• i.e. can find the data, but there seems to be some processing needed for coordinates to display properly with Datashader & HoloViews

Can we add some sort of Jupyter notebook or Python file for downloading & processing Taxi data?

• Maybe start with 2015?