HW11-Proposal_HyoSungKim #17
Conversation
|
|
||
| ## Anticipated difficulties | ||
|
|
||
| The biggest challenge for this is the amount of data points to display on the map. My current data has over 300,000 points considered possible non-owner occupied rentals and I think that the processing for this will be very slow. A possible work around this issue is to narrow down on what I plan to display, for example, filtering out and displaying rentals with risk scores greater than x. I am still considering whether or not I will by default, filter the dataset or allow the user to input a range of risk scores to filter. |
There was a problem hiding this comment.
You're exactly right that this is the central difficulty for an application of this kind. 300,000 points may well be too many, but you could perhaps aggregate at lower zoom levels and only display "full data" when zoomed in enough to naturally display fewer elements. This might be a candidate for using vector tiles (which we won't be covering in any detail I'm afraid, but which I'd be happy to give you pointers on if it comes to that)
|
|
||
| ## Missing pieces | ||
|
|
||
| It would be useful to know if there was a way to continuously update a data file with inputs and although these wouldn't be used for the visual aspects of the application, it can serve to add data into the primary dataset and as next steps, use this information to improve the model. By updating the model every now and then, risk score predictions may improve. |
There was a problem hiding this comment.
The only option for something like this would be a bit of back-end code that (nightly? hourly?) churns through inputs to update the model. The good news is that you can avoid that problem entirely for the sake of the final. If you need help pursuing next steps after that, let me know and I'll point you in the right direction
No description provided.