Exploration of SFO's the Landing data and investigating the preferred aircrafts for various types of flights.
Introduction
Working in the aviation supply chain, various aspects of aviation and air traffic can, not only be fascinating but also be impactful on those operating within the sector. A plethora of organisations offer data on Air traffic which can cover information from passenger numbers to the types of aircraft passing through the a specific airport.
Once such dataset is the Air traffic landing data from the San Francisco International Airport (SFO) data on Landings Statistics issued via City and County of San Francisco. The dataset covers a number of various factors including
Activity Period
Operating Airline
Published Airline
GEO Summary
GEO Region
Landing Aircraft Type
Aircraft Body Type
Aircraft Manufacturer
Aircraft Model
Aircraft Version
Landing Count
Total Landed Weight
Subsequently, this project will explore the data with the aim of ascertain information about the preference of certain aircrafts for specific routes and the prominence of various aircraft manufactures.
Overview
The dataset consists of over 26000 entries. However, there are some missing data. Aspects such as IATA Code is not a major issue as the key information related to this variable is stored in the operating and published airline variables.
As for Aircraft version, options to retrieve the data may be limited. It may be that for some aircraft there is simply no version.
The dataset covers the period between 2005 and 2021.
Furthermore, based on total landed weight, it appears that there has bene some fluctuations in the data. Regardless, information for 89 aircraft versions is provided in this dataset, with some version extremely popular compared to others.
Trends
Below is a graph of how the total weight of passenger, freighter and combi flights have changed over the 16 years.
Data for combination flights appear to disappear from 2015. As commercial flights can carry cargo, it may be a change in classification. However, based solely on this data, this inference cannot be validated.
The landed weight also highlights the impact of COVID which appear to have affected passenger flights disproportionally. Cargo (freighter) flights appear to have be stable with the pattern of growth from 2015 onwards appearing to be relatively unchanged. Resampling the data so as to ascertain the yearly average total landed weight. this highlights the impact of COVID in greater clarity. This graph also indicates that contrary to first observation COVID did impact freight as the growth between 2017 and 2019 appear to have been damped. presumably due to factory closures due to lockdowns.
A 3 month rolling average figures also indicate the seasonality of passenger travel, though for freight this pattern is less pronounced with more fluctuations. Also there are major spikes in the landed weigh of fright flights very few years.
 |  |
Examining the trends of total landed weight, with respect to aircraft type also indicates interesting patterns. Narrow bodies have steady increased in popularity since 2006. However, it was significantly affected by COVID and even in 2021/2022, appears to have regained its popularity.
On the other hand, wide body plans appear to have decreased in popularity over the years but it was the least affected by COVID. Wide body planes also appear to have received to pre-COVID levels in 2021/2022. COVID impact was narrow body planes appear to have a higher total landed weight.
Regional Jets appear to have disappeared from the dataset since 2015. This could be for a number of factors including the SFO no longer accepting such aircraft or they were decommissioned.
Lastly the prominence of Turbo Prop appear to be have peaked in 2014. since then it has had some fluctuations and similar to narrow body planes were significantly affected by COVID. However it has since recovered to pre-COVID levels.
The trends in aircraft type for passenger flights very much mirrors the overall trends.
Alas, for freight the data is somewhat unclear as the fluctuations in aircraft type is erratic especially for narrow body planes. As for wide body, the trend is relatively stable and highlights the growth from 2014 onwards.
Commercial/ Passenger flights
In this dataset, the majority of the data are related to passengers/commercial flights traveling both domestically and internationally. As indicated previously, commercial flights can carry some level of cargo and can be a more economical option of transporting goods. As such, the role of pure cargo flights is limited.
The airline (published and operational) data are too complex to ascertain notable information on its own, with over 85 different airlines mentioned in published airlines alone. Additionally, specific airlines appear more than others who can appear as low as only once.
Examining the GEO region for passenger flights , most flights appear to be domestic flights followed by those to Asia and Europe. However, it should be noted that neighbouring North American nations also appear to be prominent destinations, though they appear as individual countries whilst other regions cover a number of different nations. Updating the values so that US, Canada and Mexico appear as once variable highlights the prominence of northern American nations.
The location is significant as specific aircrafts are designed to cover different journeys . For example "Widebodies are exceptionally common on long-haul flights" and whilst narrow body planes can service long hauls they "are smaller" (Singh, 2019) . Therefore based upon this, one assumption may be that for passenger flights narrow body planes may be common.
This is supported in the following graphic which illustrates the dominance of narrow body planes for passenger flights. However, the difference between narrow body and wide body is not as large as could be expected.
Reviewing the distribution of the aircraft type and GEO region, shows that for North America, the data favours narrow body flights. However, for the majority of the other destinations narrow body planes are not used at all or the ratio of their usage is limited and wide body plans are the preferred/only option.
However, when regions are factored, it is clear that narrow body aircraft is a preferred choice for short haul flights such as North and Central America. However for longer distances, the wide body planes are more prominent.
In terms of manufactures of such aircraft, the market leaders Boeing and Airbus supply the majority of the narrow and wide body planes. The number of planes by Boeing dominate the list. However, in terms of ratio of narrow body to wide body planes, Boeing appears to be a favoured choice for wide body planes whilst airbus is favoured for narrow body planes. As for Regional and Turbo Prop, they appear to be supplied by Embraer and Bombardier.
Boeing has a broad array of wide body plans with the 777, followed by 747 appearing to be a prominent model.
As for Airbus, options for narrow body planes are limited to 4 types, A320, A319, A321 and A318. The Airbus A320 and A319 are the most popular types.
Singling out the Airbus narrow body planes indicate that the A318 may no longer be in use and A319 may be phased out as it appears to be dropping in significance. A320's dominance is also emphasised in the total landed weight as it rated the highest in 2016. Alas, due to the impact of the financial downturns and the impact of COVID makes it harder to decipherer whether the A321 are gaining in momentum cover the A320s.
Taking the yearly average total landed weights however does indicate that the newer A321 is indeed becoming the preferred model for the narrow body range for Airbus.
Cargo / Freighter flights
The GEO region for for freight flights are similar to commercial flights. North America and Asia are popular destinations. However, regions such as Australia / Oceania and central America do not appear in the list for freighter flights. This could be due to lack of trade to and from these regions but its more likely that it is simply the case that SFO does not host freighter flights for theses regions and/or the commercial flights are able to transport the necessary goods without a need for a specific freighter flight.
In contrast to commercial flights, cargo flights appear to consist predominantly of wide body planes. This is likely to be due to the volume of Asian flights which are greater than for commercial flights. A breakdown of the aircraft type based on destination supports this assumption. This graph also highlights that even for short haul flights, wider body planes are a preferred option.
One assumption may be that the wider body planes can carry more cargo. However the previous examination of landed weight of the plane body types, it does not appear to support this claim.
With respect to the manufacture of the freighter planes, Boeing dominates the list for wide body planes. McDonnel Douglas and Boeing are comparative with respect to the narrow body plans for freight flights.
In terms of models, the preferred model for freights appear to be the 747. It should be noted that this plan is due to go out of production from October 2022. Whilst planes typically operate for 20 years, the data may begin to change going forward as newer models gain more prominence.
Finally, whilst there are a number of missing information for aircraft version, based on the information provided the Boeing 747 version 400F appears to be the most dominant.
Sources indicate that this is model is a specifically designed as a freighter where as the general 400 could be used for both commercial and cargo flights. Further investigation. reveals that this 747-400F is no longer active as the data for the planes end in 2016.
Whilst Boeing remains still a major supplier of wide body planes, since 2016, the B744 has been the preferred model for cargo flights. Version data is alas not available for this model. However based on total landed weight, the B772 have been gaining in prominence.
Following is the fluctuations in Boeing's wide body aircraft since 2016.
Conclusion
It should be stated that the observations made in this project are focused on SFO. Therefore, some of the observations made may differ from airport to airport. That being said, based on this dataset, it is apparent that Airbus dominate the aircraft servicing commercial flights with their A320 and A321. On the other hand, Boeing is a market leader for wide body plans and dominate the cargo flights. There are also smaller manufactures who operate in the sector, however their roles are limited.
COVID a major impact on the aviation sector as travel was put on hold and lockdowns made travel difficult. However, as vaccinations become widely used, the market appear to be recovering.
Appendix
Below are a collection of visualisations highlighting the distribution of the data within a given variable.
The data for 2005 is limited and appear not to include a data for a full 12 month period. the number of flights recorded in this dataset typically ranges from 1500 to 2000 for a given year but from 2020 the number full to approx. 1250.
Whilst a overview suggest that the operating and published airlines are similar, for approx. 7.8% of the data, they differ. The cause of this difference is likely to be due to code sharing (Smithson, 2019).
It should be noted that there is a questionable entry for some domestic flights. Although referred to as domestic, there are some entries where the GEO region is listed as being Europe and international flights with a GEO region of US. For the purpose of this project, however, this is not likely to cause issues as GEO summary is not a point of interest.
Overall, US is a key GEO region in this data followed by Asia. This is a key feature of SFO. Other airports such as those based in Europe and Asia will have different results.
As one may expect, Boeing and Airbus are the main plays in the aircraft manufacturing. After these two players, there is a drop before Bombardier. Aircraft model and versions vary significantly and can only be understood in the context of the respective manufacture.
Alas, further domain knowledge is required to fully understanding the Landing Count variable. However, it appears that the data has a strong skew to the right and a broad range.
Total landed weight also has a similar distribution of as landing count with a skew to the right and a broad range of values. The range appears to be particularly broad for domestic flights.
Given that these two variables are the only numerical data in this dataset, the interaction between the two variables may reveal some insights. It appears that the values radiate from the origin and fan out in a liner manner at differing angles. However, adding GEO summary or Aircraft type both fail to account for this pattern of distribution.
 |  |
Conducting a K-Mean's cluster analyses on the scaled version of total landed weights and landing count, reveals there may be 3 -6 clusters in the data. Examining both 3 and 6 clusters reveals that the clusters are these clusters form in expanding bands from the centre. However, without further domain knowledge, understanding these patterns and clusters.
“Books are the plane, and the train, and the road. They are the destination, and the journey. They are home.” ― Anna Quindlen, How Reading Changed My Life
Comments