zhengqunkoo / taxibros Goto Github PK

Similar to gofmt.

I'll fix it later

Index by timestamp, query coordinates of all taxis in that timestamp.

Duplicates still exist in db despite single instance of running daemon.

Embed Google Maps, also include #4

We need the taxi-availability data for the model.

Sigma >1 "dissipates" heatmap, increasing number of non-zero elements in sparse matrix. So there are more elements in coo.data.tolist() than in xedges or yedges.

User moves javascript slider to scan through time. Heatmaps will change with time.

Old circle should disappear when new circle appears.

Reduce server load.

Mobile users can query heatmap times with greater control.

Form augments slider: form queries heatmap of time X, but also resets slider value to time X, and slider nub to middle.

Reduce lag of having 4000 SVGs on the map.

get_closest_roads filters coordinates within 500m before finding the roadIDs of the closest roads. This is done for coordinates of the latest timestamp, which means that the roadIDs should have already been stored.

Best way would be to compare against previous minute and see if there was a taxi that was at that location

Show the heatmap on a interactive map. User will be able to pan, zoom, rotate, query directions, and use other Google Map features.

Thus, it looks like a dashboard. And it makes it easy and more beautiful to navigate the map.

Deleting all tasks ensures only one start_download background task.
Taxibros may have other non-start_download background tasks, and these would be deleted too.
Use a less destructive method to ensure only one start_download background task.

Users can then see the date and time they query without doing mental arithmetic.

Heatmaps transferred as images. If not efficient, could transfer coordinates to client, then run heatmap conversion and overlay client-side.

chart of 24h distribution not displaying since last week. Will figure it out after I am done with nearest roads suggestion feature.

Add datetimepicker specify when free to hail taxis.
Allow users to delete locations.

User might be willing to walk longer / interested in wider area distribution.

Two separate palettes: night and day. Choose colors that invoke trust.

• Disconnect polyline between multiple search locations, if search from a different searchBox.
• Replace polylineArray if genLoc from genTimeSliderChange / showNearby / from the same searchBox.
• Disappear on itinerary table row deletion.

Add taxi count attribute to Timestamp model.
Use heatmap from #1 to filter peak locations by intensity of heat-tiles.

First step is to define a function that filters a query set of taxis at current time for taxis within a 500m radius.

Next step is to display this on the map.

Finally, extend this across time to generate the distribution.

An itinerary: a (not necessarily continuous) sequence of journeys.
A journey: users walking from their current location to a recommended taxi hailing location, then riding a taxi to their desired destination.

Target audience: anyone planning a journey in advance, where some parts of the journey is by taxis. Mostly, but not exclusively tourists.

Inputs:

• 1. Locations where they want to take a taxi.
• 2. Times when they are free to take a taxi.
• 3. Their desired destinations.
• 4. Arrival times at their desired destinations.

Outputs:

• 1. Recommended hailing location. See issue #35.
• 2. The time users reach the recommended hailing location (i.e. time taken to walk from Input1 to Output1).
• 3. Whether the journey is feasible (i.e. whether taxi journey time exceeds time duration between Output2 and Input4).

Assume most users use mobile, so use phones to access our website.

Ask them to perform two tasks, then ask questions after each task:

1. Find nearby taxis. Are they willing to walk there to hail the taxi? Would they instead pay the booking fee for the convenience of not walking there?
2. Plan an itinerary: a (not necessarily continuous) sequence of journeys. Each journey is: walking from their current location to a recommended taxi hailing location, then riding a taxi to their desired destination. Do they trust our app enough to follow our suggestions for the entire trip? What if this trip is in a foreign country, would they be less trusting?

Create a function to calculate taxi fare based on distance.

• Location filter should filter on the taxi distribution at the time specified in the slider, instead of the latest timestamp.
• Time filter should filter on the few taxis given by the location filter.

Export for storing itinerary in user device, independent of taxibros.

Exported data should be human-readable.
Importing data should have same effect as manually entering data.
See Outputs and Inputs sections of issue #37 for more info.

Should sort by user place name to avoid confusion.

User can then plan itinerary by seeing taxi availability in area, across time.

Might be time to write tests?

Some simple ones at least
https://www.softwaretestinghelp.com/web-application-testing/

I believe its because the script to load the map is called before the DOM element is created. Hence the reference error. (Look at bottom right corner)

On Ubuntu 16.04.4 LTS: FileNotFoundError: [Errno 2] No such file or directory: '/proc/10837/stat'

Wrong number of rows, missing icons.

It is hard for a user to get information from discrete points. Heatmaps clump coordinates into regions, and add color for intuitive visuals.

Pass ?date_time= to URL request. Gets list of date_times not already in database. Adds date_time to database only if it is not already in database.

Therefore, variables are not properly accessed by django

Let users search their location of interest.

Query genLoc using coordinates of searched location (instead of geolocated location).