
import Iframe from "./maps/Iframe";
import AhmedHelmy_Badr_MapHtml from "./maps/Ahmed Helmy Station-Badr City.html";
import Abasseya_HadabaWosta_MapHtml from "./maps/Abbasseya Station-Hadaba Wosta (Moqattam).html";

Formal transit networks are quite predictable and well-defined. Paratransit modes, on the other hand, can exhibit dynamic, interesting patterns that are unique to each local context. Identifying such patterns requires extensive data that covers the majority of trips and modes in the targeted region.

This analysis leverages TfC's GCR mobility data gathered over multiple data collection surveys, which were orchestrated using TfC's RouteLab Software Suite.

## Data Overview

We start by reading in all the trips into a dataframe.

<div
  style={{
    fontFamily: "monospace",
    overflow: "scroll",
    width: "100%",
  }}
>
  <table border="1">
    <thead>
      <tr style={{ textAlign: "right" }}>
        <th></th>
        <th>gid</th>
        <th>trip_short</th>
        <th>agency_id</th>
        <th>vehicle_name</th>
        <th>len_km</th>
        <th>fare</th>
        <th>geom</th>
      </tr>
    </thead>
    <tbody>
      <tr>
        <th>0</th>
        <td>3025</td>
        <td>Madinat Al Solb (Al Tibeen)-Ramses Station</td>
        <td>CTA</td>
        <td>Bus</td>
        <td>36.885011</td>
        <td>NaN</td>
        <td>LINESTRING (1116749.522 1183560.929, 1116734.6...</td>
      </tr>
      <tr>
        <th>1</th>
        <td>3026</td>
        <td>Bashtil-1st Settlement Station (New Cairo)</td>
        <td>CTA</td>
        <td>Bus</td>
        <td>41.396419</td>
        <td>NaN</td>
        <td>LINESTRING (1103876.199 1216613.425, 1103865.9...</td>
      </tr>
      <tr>
        <th>2</th>
        <td>3027</td>
        <td>Athar Al Naby-Khosous</td>
        <td>CTA_M</td>
        <td>Minibus</td>
        <td>32.482129</td>
        <td>NaN</td>
        <td>LINESTRING (1108654.344 1206291.595, 1108667.7...</td>
      </tr>
      <tr>
        <th>3</th>
        <td>3028</td>
        <td>Al Omraneyya Al Gadida-Ibrahim Bek</td>
        <td>CTA</td>
        <td>Bus</td>
        <td>24.617619</td>
        <td>NaN</td>
        <td>LINESTRING (1104342.057 1206034.178, 1104337.0...</td>
      </tr>
      <tr>
        <th>4</th>
        <td>3029</td>
        <td>Athar Al Naby-8th District (Nasr City) - Al tabba</td>
        <td>CTA</td>
        <td>Bus</td>
        <td>26.152755</td>
        <td>NaN</td>
        <td>LINESTRING (1108653.298 1206291.112, 1108667.7...</td>
      </tr>
    </tbody>
  </table>
</div>

## Number Of Trips Per Mode

The `vehicle_name` field is the one of interest. Let's see what types of vehicles are used across the trips.

<FullWidthComponent>![](./imgs/blog_post_13_0.png)</FullWidthComponent>
Quite interesting already. Microbuses are dominating the scene! Buses come in second,
together with microbuses taking on more than two thirds of the total number of trips.

Below is a table showing how these modes different vehicles look like, their varying passenger capacities, licensing, and operators are all factors affecting their operational patterns.

|          |                          |
| -------- | ------------------------ |
| Tomnaya  | ![](./imgs/suzuki.jpg)   |
| Microbus | ![](./imgs/microbus.jpg) |
| Box      | ![](imgs/box.png)        |
| Bus      | ![](imgs/bus.jpg)        |
| Minibus  | ![](imgs/minibus.jpg)    |

## Fare Policy

Let's then explore the fare system for each mode. We plot the trip fare against trip length, disaggregated by `vehicle_name`.

![](./imgs/blog_post_19_0.png)

These plots tell us a lot about differences between the modes:

Bus & Minibus:

- Buses and Minibuses cover quite a large range, still smaller than that of the Microbus, from less than 5 to ~15 km.

Microbus:

- Microbuses cover the largest range of trip lengths. i.e. the most versatile vehicle, used for local and express transit services.

- Stratified fare without regard to length of trip. The fares histogram is dominated by the 5 EGP trips.

- Microbus fares have a clear linear relationship with the trip length. That's due to the market dynamics setting the prices for the microbus services. Tomnaya trips exhibit the same nature.

Tomnaya:

- The Tomnaya's covers a low range of trip lengths, showing that it primarily serves last and first mile trips.

## Speed

Averaged stats over a whole trip are not good enough for analysing the vehicle speed accurately. This section requires more granular data; at the level of segments stretching between pairs of stops.

Each two consecutive stops are considered an origin-destination (OD) pair. Average speed between stops is calculated for each mode and for each time interval across the day, using raw GPS data from multiple onboard surveys conducted by TfC.

We plot the histogram of speed per vehicle type.

![](./imgs/blog_post_25_0.png)

![](./imgs/blog_post_26_0.png)

Similar to previous observations, the Bus and Minibus show the most similar patterns, with a mean at ~20 km/hr.

Microbus and Tomnaya have a similar mean and Inter-Quartile Range. However, the microbus's range is the uniquely the largest. Its distribution is also the most leaning towards higher speeds. Apart from having a smaller weight and passenger capacity, the higher means for th emicrobus and tomnaya show that they face less congestion than the other modes. The tomnaya primarily serves first and last mile rides in sprwaling new urban communities; where residential area roads are built wide and experience moderate demand. The microbus' high speeds, however,arise in long-distance express trips.

## Boarding and Alighting (BA) Patterns

The (Fare vs. Trip Length) plots show that the main PT vehicles, bus, minibus and microbus, are all utilized to cover a large range of trip lengths, with the microbus trips having the largest range. However, visualizing boarding and alighting patterns along shared trip routes can help us better understand more nuanced and geographic differences between those modes.

In this part of the analysis, we actually need to dive deeper into the raw trip traces collected from the field. This raw field research data will show us the actual recorded BA patterns.

The most dissimilar types of PT vehicles on Egyptian streets are buses (~50 passengers) and microbuses (14 passengers). This is most apparent when comparing BA patterns. Moreover, manual inspection of trip GPS traces calls for a smaller, more manageable subset.

![](./imgs/blog_post_34_0.png)

Keeping in mind that these are average trip lengths, we can observe that for the same Origin and Destination Terminals, a microbus is usually faster than a bus, even though the route length using both vehicles may be very close. `Ahmed Helmy Station-Badr City` looks quite interesting; bus/microbus trip lengths are almost equal but trip duration duration by microbus is half the duration by bus.

`Abbasseya Station-Hadaba Wosta (Moqattam)` and `Al Qanatir Al Khayriyyah-Abd Al Moneim Riad Station` trips are also have a significant difference between trip durations by bus and microbus. However, also the trip lengths differ in the former, which raises a question about the chosen route by the agency that runs each type of vehicle.

`Ahmed Helmy Station-Badr City` and `Abbasseya Station-Hadaba Wosta (Moqattam)` are, therefore, suitable examples for further investigation of other aspects, such as Boarding/Alighting data.

Let us start with the trip `Ahmed Helmy Station-Badr City`. We plot the recorded bus and microbus onboarding instances of the trip in an interactive map.

<Iframe
  srcDoc={AhmedHelmy_Badr_MapHtml}
  width="100%"
  height="300"
  frameBorder="0"
/>
<br />
<em> Click on a stop to view board/alight numbers.</em>

Given that this is the longest trip among the bus/microbus trips (50 km) and given the distribution of stops along the trip, it clearly is more of an express trip. There are only 3 stops in the microbus trip, compared to more than 15 stops in the bus trip.

Microbuses follow a fill-and-go system. Long-distance microbus trips do not board on the road only if a passenger alighted, which is uncommon. The microbus fare is usually higher than the bus fare for express trips, few passengers would pay the higher fare to alight mid-trip. Most passengers take the whole trip from origin to destination.

The recorded bus trips, which take pretty much the same route, are quite different. The bus moves from the origin terminal on time, regardless of number of boarded passengers. For the first half of the trip, few passengers alight from the bus, but much more board. Throughout the rest of the trip, boarding decreases significantly, alighting continues, but most passengers still alight the bus at the destination terminal.

The bus BA patterns show that it operates primarily as an express service, but still serves local trips along its route, contrary to the purely express trips offered by the microbus.

The BA patterns can explain the large difference in this trip's duration between bus and microbus, as shown in the bar plot. The bus stops at many points to board or alight, accumulating delay, as opposed to the microbus taking a one-shot trip to the destination.

Pretty insightful so far! Let's explore the second example trip `Abbasseya Station-Hadaba Wosta (Moqattam)`.

<Iframe
  srcDoc={Abasseya_HadabaWosta_MapHtml}
  width="100%"
  height="300"
  frameBorder="0"
/>
<br />
<em> Click on a stop to view board/alight numbers.</em>

This is a short trip, compared to the previous one. Still, it shows an even larger difference between bus and microbus in trip duration. The microbus takes a much shorter route than that of the bus.

The microbus BA pattern resemble that in the first trip, where passengers board at the beginning and only start to alight towards the end of the trip. This explains the route choice for the microbus; it takes the shortest route to the destination.

On the other hand, the bus takes a route that seems too long. The board/alight numbers justify that choice, though. This bus does not only operate as an express trip but serves local trips demand as well.

# Conclusion

The bottomline from this BA patterns analysis is that express microbus trips are faster than their bus counterparts. while usually being higher in fare, and not having a reliable headway.
On the other hand, express bus trips generally have fixed headway and a lower fare, but are delayed by frequent stops along the way.


Public transit in the Greater Cairo Region (GCR) exhibits notable diversity, where formal and informal services serve millions of passengers across many different modes and routes. Big picture questions arise from such a complex environment— How are fares distributed? Which modes are more popular? and other questions. In this post, we dive into TfC's Cairo dataset and find out interesting answers with statistical analysis and visualizations.

Comparing Public Transit Modes in Greater Cairo


## The Challenge

Movement in Kampala, Uganda’s capital and home to [1.5 million citizens](https://sustainabledevelopment.un.org/content/documents/3759kayemba.pdf), is no simple [chore](https://www.youtube.com/watch?v=cjfW_HX9WtM&t=1448s&ab_channel=HOPEConsults%28U%29Ltd).
On an average day, [two thirds](https://sustainabledevelopment.un.org/content/documents/3759kayemba.pdf) of trips are completed by privately operated taxis or Matatus; while the majority of the remaining trips are provided by motorcycles or “Boda Bodas”. The former are 14 seater microbuses, and the latter are motorcycles operating privately. At the time of the project, none of these modes were regulated. In addition, of the 2,100km road network on which they operate, only 500kms were paved. Even though the Kampala Capital City Authority is currently paving its roads, the lack of transport regulation resulted in an over-supply in the paratransit
sector causing daily traffic congestion, [business losses](https://www.theigc.org/project/the-cost-of-congestion-in-kampala-uganda/) and immobility all over the city. Over the course of a year, Transport for Cairo has been working on a paratransit and street usage study in preparation for the  Bus-Rapid Transit (BRT) pilot project system. The BRT system is envisioned by the government to help alleviate congestion in the Greater Kampala Metropolitan Area (GKMA).

![Old taxi Park in Uganda’s Capital Kampala. Credit Wambi Micheal /IPS](./imgs/Wambi.jpg "Old taxi Park in Uganda’s Capital Kampala. Credit Wambi Micheal /IPS")

## The Process

For 10 weeks, the TfC team mapped the public transport network, and provided the operational data necessary to understand the paratransit services in the GKMA area. Through Route Observer (RO), a mobile app developed by TfC, we collected a total of 3200 hours of data points on 1492 trips and 587 routes all over the GKMA. The first steps included using both the data collected during the pilot mapping and secondary sources in order to identify origin and destination points.

![Route Observer Application: The first steps in choosing an onboard assignment](./imgs/app.png "Route Observer Application: The first steps in choosing an onboard assignment")

Then a standardized list of terminals was periodically uploaded to our Route Observer Dashboard application in order to create assignments for the field researchers. When a field researcher picks an assignment and starts recording on the RO app, the app records both the GPS trace
of the trip as well as travel time.Then using Route Snapper, a software developed in-house by TfC,
incoming raw traces are processed in batches and migrated to the
“Processed Trips” layer in the database.

![Raw Trace (Dotted Blue) overlaid with its processed trace aligned with OSM (Red)](./imgs/raw-traces.png "Raw Trace (Dotted Blue) overlaid with its processed trace aligned with OSM (Red)")

After data collection, the team created a virtual network of stops in order to model the transportation network. This dataset contributes to the creation of a GTFS with estimated travel times (GTFS feeds are used for trip planning and transport network analysis to evaluate multimodal transport options). Finally, we created a Kampala Bus Map designed to follow the geographic arrangement of the city, while abstracting and simplifying it. The streets shown on the map are simple to read and understand. They are divided into four categories based on street hierarchy, with each category having a different opacity level based on the street’s usage importance. Only the most important streets, defined by the frequency of usage as well as the streets on which the trips are placed, are shown on the map.

    ![Kampala Transit Map Progress](./imgs/final-map-1536x513.jpg "Kampala Transit Map Progress")

## Today

Because we know the value of capacity building, we partnered up with World Resource Institute (WRI) and conducted an OpenStreetMap workshop (Mapathon) last January at Makarere University, College of Engineering Design Art and Technology. We worked to raise awareness towards the need for a data management system that can be maintained by locals in order to achieve the most sustainable results in the long run. In collaboration with WRI, the aim of the workshop was to build capacity for students and fresh graduates working with Open Data and OpenStreetMap and to raise awareness on the impact of such platforms and concepts on society. Presenters from Addis Ababa Transport Authority and Accra Ghana talked about their experience in independent data creation and storage. The workshop also incentivized participants to become contributors and active members of their relevant local mapping communities, owning the process of creating and maintaining the data.

![Abdelrahman Meligy, Project Manager, giving Makarere University students one of the workshop sessions](./imgs/workshop.jpg "Abdelrahman Meligy, Project Manager, giving Makarere University students one of the workshop sessions")


In Kampala, Uganda, citizens rely completely on informal transportation, in the form of matatus or boda bodas, for all their transportation needs. This reliance, coupled with a lack of regulation, has resulted in over-supply in the paratransit sector. Transport for Cairo was able to complete 10 weeks’ worth of data collection using our Route Observer application. We observed 1492 trips and 587 routes operating out of Matatu stages across GKMA. Video feeds were used for traffic counts and machine learning algorithms were used to classify vehicles across selected sections for analysis.

Kampala Paratransit Street Usage Study

In transport planning, it is often useful to zoom in and understand how travel patterns map onto the road network. This post offers a detailed walkthrough of how to determine the number of (public transport) vehicles passing on each road in a city. We take Cairo as a case study, and identify the roads that are most heavily utilized by public transport.