mibici in Guadalajara
There are many ways to evaluate policy. This isn’t a paper, perhaps user adoption is one of them, let’s take a look. My hometown, Guadalajara adopted one of those bike sharing programs some years ago. It’s called mibici (mybike). They have some open user data and based of what I’ve seen, it can be quite popular:
Figure 1: Total trips.
The plot above is most of the trips that they have saved. There’s a lot of observations missing in 2017. They changed the format for some strange reason, some encoding was lost, and all of those trips had to be dropped with extreme prejudice. Such is life. Guadalajara, a non-tropical city in the tropics with basically 2 seasons (wet and dry) looks like a giant grid of cement.
Figure 2: It’s lovely.
As fun and interesting as a giant grid of cement looks from above (1 degree is around 69 miles/111 km), it just won’t work with those colors. Another map will have to do:
Figure 3: Much better.
Coverage and usage
This is the absolute current coverage of the program (or at least, until the end of 2018). Each dot represents one station, regardless of whether they are active, inactive, or how many bikes they have. The highest concentration is in the actual downtown of the city (in general, the historic center will host the cathedral, as is the case here. Thankfully the map shows it as a somehow obscure point of interest).
Figure 4: Current total stations.
However, it is important to notice that the program did not start with all of these stations but rather, new were added in time. mibici started mostly downtown and gradually added more stations, not only around the same area, but also in new locations (some sort of type), here’s the station growth:
Two things are immediately interesting to me. First is that there is this one random station in the middle of nowhere (actually it’s somewhere around here), a pretty populated area. The thing is, this is the only station of this group anyway, either active or inactive. Second, it seems that some stations were shuffled around, in very similar locations. No explanation is (was) given for this in their site.
| location | total |
|---|---|
| POLÍGONO CENTRAL | 213 |
| ZAPOPAN CENTRO | 40 |
| TLQ-CORREDORATLAS | 32 |
| POLÍGONO NUEVO | 1 |
Below is a timelapse of the usage of the stations by gender where applicable, since the beginning of the program until the end of 2018. It seems that the designers did a good job, as the biggest circles are in the ‘central polygon’ (POLÍGONO CENTRAL, also the group that has the most stations), more trips originating from downtown than the periphery.
The arrivals below, are not much different. For trips that have a median duration of less than 10 minutes (in my own personal experience, around a mile -1.6 km), I’d assume they’re trying to beat traffic around the same zone.
It seems that gender has bad encoding (actually, there’s a lot of bad trips logged, hundreds of thousands are classified as NA and were removed originally), for some months in 2017, it seems that only men used thed service, which may not necessarily be the case. A bunch of non classified gender trips (also not classified as NA) could be thought of as belonging to females as the proportions for the rest of the trips may seem to match. However, this is only a supposition, and there’s no other information that can corroborate. Below the potential trips taken by gender:
Figure 5: Trips by Gender
Total usage
Figure 6: Total usage by the end of 2018
Even with all of those trips from women removed, it does seem that the usage is very similar among groups. Here’s the box plot of trip duration by seconds, the median usage is below 10 minutes (600 seconds). The average, due to the tremendous amount of outliers (1,319,981 trips were removed, wow) is 1237 seconds (slightly over 20 minutes, more than double). The ones on the right are violin plots, which show the density. Uhhh… I’m not even going to comment on those shapes. Althought the pink-ish one (F) kinda looks like a thinned out, whatever this is, from that Deftones album.
Figure 7: Box/Violin plots by gender (outliers removed)
The committment to the cause or, crime regret
One of the more interesting aspects of the data as its currently saved, is that on the one hand, people seem to love this stuff. That gifs above show that for some stations, daily outgoing trips are above 200. I’m not so sure how the bikes get re-arranged during the day, either by the users themselves or by people that work for the program, but the daily activity looks pretty ‘healthy’ (for lack of a better word). Another interesting result, that can be implied, is that they simply are not able to tell (easily) if a bike was stoline. The pricing model is per day/year rather than minutes used. I cannot see any penalties (at least, not easily) for returning the bikes late. Take a look at the below, some people (genders withheld to protect the non-innocent) took the bikes for months, presumable built close relationships with them, and for some strange reason decided to return them:
| user | start of trip | end of trip | duration | days |
|---|---|---|---|---|
| 13296 | 2016-10-04 20:55:51 | 2017-06-21 15:43:46 | 22445275 secs | 259 |
| 12091 | 2016-10-11 13:56:33 | 2017-05-28 22:51:51 | 19817718 secs | 229 |
| 13841 | 2016-10-06 08:07:39 | 2017-05-22 09:11:36 | 19703037 secs | 228 |
| 14142 | 2016-10-17 14:54:29 | 2017-06-01 09:06:05 | 19591896 secs | 226 |
| 13450 | 2016-10-04 12:48:44 | 2017-05-17 18:02:31 | 19458827 secs | 225 |
| 816 | 2016-11-03 14:03:02 | 2017-06-16 13:37:36 | 19434874 secs | 224 |
| 3495 | 2016-10-25 16:59:32 | 2017-06-07 13:32:52 | 19427600 secs | 224 |
| 15517 | 2016-11-17 11:36:03 | 2017-06-25 16:39:29 | 19022606 secs | 220 |
| 14948 | 2016-11-10 21:16:37 | 2017-06-18 12:39:09 | 18973352 secs | 219 |
| 12472 | 2016-11-04 09:20:17 | 2017-06-08 10:47:21 | 18664024 secs | 216 |
More than that, there are at leas 750 trips that took at least one 1 day (where are they biking? The place is mostly flat) or more. It makes you really wonder how many of these bikes have been written off, and then just ‘came back to life’.
On the other hand, there seems to be a lot regret trips: 7,247 trips lasted 0 seconds. This does not count the first 2 recorded trips (December 1st 2014), by user 1436 which were round trips (see below) and lasted fewer than 5 minutes each, at midnight, when the following trips started after 9:30 that same date.
The actual round trip or, bike regret
Perhaps some of the more interesting results (at least for me) are the actual round trips (not counting these that lasted 0 seconds above, which were by some definition round). 434,534 trips start and end in the same station. What are these people doing? Where are they going (aside from… the exact same place). Maybe they regretted taking a bike in the first place. Nothing else to see here, I just thought it was interesting.
During the week
The work week in Guadalajara (and in Mexico, in general), is Monday to Friday, and there is a lot of retail downtown. I find surprising that there seem to be fewer trips during the weekened. Maybe people are not using this for recreation, but rather for commuting purposes. The public transportation network is quite defficient, I would see why people woud rather bike if they can and… If they don’t mind the heat.
What now?
No idea, any thoughts? Maybe trying to see the population density, and try to figure out how exactly are the bikes being re-arranged during the day, how do they choose new places for stations?
