Dhanujie Jayapala

Outline

• Executive Summary
• Introduction
• Methodology
• Results
• Conclusion

2
 Executive Summary
• Summary of methodologies
In this exercise team used multiple methodologies to acquire and analyze the causal relationship of
successful rocket landings by Space X.
• Collect : Data collected using SpaceX REST API and web scraping techniques acquired data from Wikipedia
• Wrangle : Convert raw data to create usable outcome variable
• Explore : Data visualization techniques to explore trends, considering factors like payload, launch site, yearly
trends.
• Analyze : Analyzing the data with SQLlite
• Geographic Mapping : Geographically visualize the launch site success rates and proximity to geographical locations
using folium
• Dashboard : An interactive dashboard website launch sites with the most success and successful payload ranges
• Build Models : Machine Learning to predict landing outcomes using logistic regression, support vector machine
(SVM), decision tree and K-nearest neighbor (KNN)

3
 Executive Summary

• Results can be Summarized into three buckets.

• Exploratory Analysis
• Success of missions has increase with time
• We can observe a 100% success rate for ES-L1, GEO , HEO and SSO orbit types
• There were serval drone ship landings

• Visualization
• Highest performing payload mass is between 2,000 and 6,000 kg
• Space X has chosen sites near the equator. Other factors such as close to a main road and railway was also import
• Being close to a coast makes some room for failed landings
• Most successful site is KSC LC-39A with a success rate of 77

• Predictive Analytics
• Decision Tree Classifier has the best learning algorithm for the data available
4
 Introduction
• SpaceX, officially known as Space Exploration Technologies Corp., is a
revolutionary American aerospace company founded by Elon Musk in
2002.
• Key Accomplishments:
• Reusable Rockets: SpaceX has successfully developed and implemented reusable rockets like the Falcon 9
and Falcon Heavy, significantly reducing launch costs.
• Human Spaceflight: SpaceX achieved the milestone of sending astronauts to the International Space
Station (ISS) with its Dragon spacecraft, marking the first time a private company has transported humans
to orbit.

• Impact on the Space Industry:

• Reduced Launch Costs: SpaceX's reusable rockets have dramatically lowered the cost of accessing space,
opening up new possibilities for commercial and scientific endeavors.
• Increased Innovation: The company's focus on innovation and pushing technological boundaries has
spurred advancements in various areas of space technology.

• Future Outlook:
• SpaceX continues to drive the future of space exploration with its
ambitious projects like Starship and Starlink. The company's
innovative approach and commitment to reducing costs have
positioned it as a major player in the global space industry, shaping
the future of space travel and exploration.

5
 Objectives of the
Analysis
• Understand key considerations
for the success of Space X

• Identify factors contributed to

the success of the Space X
program

• Learn improvement patterns

from the data available on
Space X

• Replicate the findings for

better success for the in Strat
up Space Y 6
Section 1

7
Methodology

Executive Summary
• Data collection methodology:
• Data were collecting mainly using two methodologies, which are web-scarping through
Wikipedia and using SpaceX REST API platforms.

• Perform data wrangling

• Data preparation or wrangling was done by filtering the dataset, removing missing
values and doing one hot encoding.

• Perform exploratory data analysis (EDA) using visualization and SQL

• Thr

• Perform interactive visual analytics using Folium and Plotly Dash

8
Data Collection

Identify the REST Drop or Filll NA

Export Data to CSV
API supported URL values

Call the GET Filter from the

request Dataframe

Convert the JSON

Check if the status
response by Pandas
code is 200
lib to a Dataframe
9
 Data Collection – SpaceX API

• Key Code lines

Import requests and Filter from the
Name the columns
Pandas libraries Dataframe
• # Requests allows us to make HTTP requests which we will use to
get data from an API
• import requests
• # Pandas is a software library written for the Python
programming language for data manipulation and analysis. Convert the JSON
Identify the REST Drop or Filll NA
•
response by Pandas
import pandas as pd API supported URL values
lib to a Dataframe
• response = requests.get(spacex_url)

• data = pd.json_normalize(response.json())

• data.head() Check if the status

Call the GET request Export Data to CSV
code is 200

Full Code Can be found in here: 10

 Data Collection - Scraping
• Key lines of Code
import requests
Import
from bs4 import BeautifulSoup Find the table
Beautifusoup4 and Extract column
contents through
request libraries names to a list
import pandas as pd libraries
FindAll(‘table’)

html_data =
requests.get(static_url)
Read html data
html_data.status_code Identify the REST Add column names
using Beautiful Soup
API supported URL to a Dataframe
Obejct
soup =
BeautifulSoup(html_data.text)
html_tables =
Write to dataframe
soup.find_all('table’) Call the GET request
Check if the status
for each row with
code is 200
FindAll tr and td

11
Full code can be found here
 Data Wrangling
• Key lines of Code
import pandas as pd
Import Pandas
import numpy as np
df=pd.read_csv(csv_file_path)
df.isnull().sum()/len(df)*100

good_outcomes = Do one hot

set(landing_outcomes.keys()).difference(bad_outc Load data from the encoding for the
omes)
CSV selected objects (
landing_class =
df['Outcome'].replace(good_outcomes, 1)
Success vs Failure)
landing_class =
pd.DataFrame(landing_class).replace(bad_outcomes
, 0)

Identify null values Identify data types

12
Full code can be found here
EDA with Data Visualization
• Charts plotted and why
• Flight Number vs Payload Mass with
Success in Blue : To see how payload mass
increased over time and success rate
increased over time
• Flight No vs Launch Site : What are more
successful launch sites and what is used
nowadays
• Payload Mass vs Launch Site : To see if
Launch site supports certain Payloads

• Full workbook link :

13
EDA with Data Visualization

• Charts plotted and why

• Orbit Type and Success Rates : To see any
relations ship to see if there are any orbits
that causes mission failures
• Payload Mass vs Orbit Type : Any specific
orbits types require a certain payload mass
• Year vs Success Rate: How identify whether
the success rate has improved over the
years

• Full workbook link :

14
EDA with SQL

• Using bullet point format, summarize the SQL queries you performed

• %sql create table SPACEXTABLE as select * from SPACEXTBL where Date is not null

• %sql SELECT DISTINCT "Launch_Site" FROM SPACEXTBL

• %sql SELECT * FROM SPACEXTBL WHERE "Launch_Site" LIKE "CCA%" LIMIT 5

• %sql DESCRIBE SPACEXTBL

• %sql SELECT AVG(PAYLOAD_MASS__KG_) FROM SPACEXTBL WHERE "Booster_Version" LIKE "F9 v1.1%"

• sql SELECT MIN(DATE) FROM SPACEXTBL WHERE "Mission_Outcome" IS "Success"

15
EDA with SQL

• Using bullet point format, summarize the SQL queries you performed
• %sql SELECT DISTINCT "Booster_Version" FROM SPACEXTBL WHERE "PAYLOAD_MASS__KG_" BETWEEN
4000 AND 6000

• %sql SELECT COUNT("Mission_Outcome") FROM SPACEXTBL WHERE "Mission_Outcome" IS "Success"

OR "Success (payload status unclear)"

• sql SELECT COUNT ("Mission_Outcome"),"Mission_Outcome" FROM SPACEXTBL GROUP BY

"Mission_Outcome"

• %sql SELECT DISTINCT "Booster_Version" FROM SPACEXTBL WHERE "PAYLOAD_MASS__KG_" IS (

SELECT MAX("PAYLOAD_MASS__KG_") FROM SPACEXTBL)
• %sql SELECT substr("Date",6,2) as "Month" , substr("Date",0,5) as Year , "Landing_Outcome", "Booster_version",
"Launch_site" FROM SPACEXTBL WHERE "Landing_Outcome" IS "Failure (drone ship)" AND SUBSTR("Date",0,5) IS "2015"

• %sql SELECT "Mission_Outcome", COUNT("Mission_Outcome") as "Events" FROM SPACEXTBL WHERE "Date" BETWEEN "2010-06-04"
AND "2017-03-20" GROUP BY "Mission_Outcome" ORDER BY "Events" DESC

• Full code is here :

16
Build an Interactive Map with Folium

• Summarize what map objects such as markers, circles, lines, etc. you created and
added to a folium map
• Explain why you added those objects
• Add the GitHub URL of your completed interactive map with Folium map, as an
external reference and peer-review purpose

17
Build a Dashboard with Plotly Dash

• Summarize what plots/graphs and interactions you have added to a

dashboard
• Explain why you added those plots and interactions
• Add the GitHub URL of your completed Plotly Dash lab, as an external
reference and peer-review purpose

18
Predictive Analysis (Classification)

• Summarize how you built, evaluated, improved, and found the best
performing classification model
• You need present your model development process using key phrases and
flowchart
• Add the GitHub URL of your completed predictive analysis lab, as an external
reference and peer-review purpose

19
Section 2
Flight Number vs. Launch Site

• Class 0 indicates failures and Class 1 indicates success.

• We can observe that there is a higher failure rate in the first few years and the failure rate as
reduced over the years.
• Also, Initially the site they operated was mostly from CCAF5 5LC 40 and later moved to K5C LC
39A which has a higher success rate.

21
Payload vs. Launch Site
We can see the Falcon X team has done very high payloads ( ~ 15,000 kg) with a
high success rate.
Most launches around 7000 kg were successful.
VAFB SL 4E site was not used for launches about 10,000 kg

22
Success Rate vs. Orbit Type
• We can observe a 100% success rate for
ES-L1, GEO , HEO and SSO orbit types

• SO has a 0% success rate.

• GTO, ISS , MEO, PO, LEO and VLEO has

an increasing success rates starting from
50% to 90% in consecutive order.

23
Flight Number vs. Orbit Type

Most initial launches are made from LEO, ISS,

PO and GTO orbits.
For more recent flights they have preferred
VLEO orbit.
The success rate has increased as the flight
no increased, meaning the recent flights with
higher success rate.

24
Payload vs. Orbit Type

• Large payloads were sent to

VLEO, ISS and PO orbits

• GTO has been attempted

with multiple medium level
payloads, with mixed results.

25
Launch Success Yearly Trend

• There were no success for

the first 4 years.
• The success rate has
improved over the years with
an exception on year 2018
since 2013

26
All Launch Site Names

• Unique launch site names are

• CCAFS LC-40
• VAFB SLC-4E
• KSC LC-39A
• CCAFS SLC-40

• The SPACEXTBL has all the launches. Hence we need to find distinct names
under “Launch Site” column to see the names of all sites.
• Query to get the above result are as below.
• %sql SELECT DISTINCT "Launch_Site" FROM SPACEXTBL

27
Launch Site Names Begin with 'CCA'

• Records starting the launch site Date

Time Booster_ Launch_
Payload
PAYLOA
D_MASS Orbit
Custome
Mission_ Landing_
Outcom Outcom
name as CCA are to the right. (UTC) Version Site
__KG_
r
e e
Dragon

• We have selected five results from 2010-06-

04
18:45:00
F9 v1.0
B0003
CCAFS
LC-40
Spacecra
ft 0 LEO SpaceX
Failure
Success (parachu
SPACEXTBL, then on Launch Site Qualifica
tion Unit
te)

column , we have search for strings Dragon

demo
starting with CCA. % denotes flight C1,
trailing string with any character/s. 2010-12-
15:43:00
F9 v1.0 CCAFS
two
CubeSats 0 LEO (ISS)
NASA
(COTS)
Failure
Success (parachu
08 B0004 LC-40
, barrel NRO te)
• The code to get the above result is Brouere
of

as follows. cheese
Dragon
2012-05- F9 v1.0 CCAFS NASA No
• %sql SELECT * FROM SPACEXTBL 22
7:44:00
B0005 LC-40
demo
flight C2
525 LEO (ISS)
(COTS)
Success
attempt
WHERE "Launch_Site" LIKE "CCA%" 2012-10- F9 v1.0 CCAFS SpaceX NASA No
0:35:00 500 LEO (ISS) Success
LIMIT 5 08 B0006 LC-40 CRS-1 (CRS) attempt
2013-03- F9 v1.0 CCAFS SpaceX NASA No
15:10:00 677 LEO (ISS) Success
01 B0007 LC-40 CRS-2 (CRS) attempt
28
Total Payload Mass

• Total Payload Mass for the customer NASA CRS missions are 45,596 kg

• Here we are selecting a sub table of PAYLOAD_MASS_KG , from the

SPACEXTBL , when the customer is NASA CRS and summing the result.

• Code is as following
• %sql SELECT SUM(PAYLOAD_MASS__KG_) FROM SPACEXTBL WHERE CUSTOMER IS
'NASA (CRS)’

29
Average Payload Mass by F9 v1.1

• The average payload mass by F9 v1.1 is 2928.4

• Here we do a small table of SPACEXTBL where the Booster version is F9 v1.1
and getting the average of the results.
• The SQL query goes as below.
• %sql SELECT AVG(PAYLOAD_MASS__KG_) FROM SPACEXTBL WHERE
"Booster_Version" IS "F9 v1.1"

30
First Successful Ground Landing Date

• First successful landing outcome on ground pad was on 22/12/2015

• The query ran was
• %sql SELECT MIN(DATE) FROM SPACEXTBL WHERE "Landing_Outcome" IS "Success (ground pad)"

• Finding the earliest data (minimum) of which Landing outcome is “Success

(ground pad”

31
Successful Drone Ship Landing with Payload between 4000 and 6000

• Boosters which have successfully landed on drone ship and had payload mass
greater than 4000 but less than 6000 Booster_Version
F9 FT B1022
F9 FT B1026
F9 FT B1021.2
• SQL query is as below. F9 FT B1031.2

• %sql SELECT DISTINCT "Booster_Version" FROM (SELECT "Booster_Version"

, "Landing_Outcome" FROM SPACEXTBL WHERE PAYLOAD_MASS__KG_ BETWEEN
4000 AND 6000) WHERE "Landing_Outcome" IS "Success (drone ship)"

32
Total Number of Successful and Failure Mission Outcomes

• Mission Outcomes are as follows.

1 Failure (in flight)

98 Success
1 Success
1 Success (payload status unclear)

• Query result is as follows

• %sql SELECT COUNT ("Mission_Outcome"),"Mission_Outcome" FROM SPACEXTBL
GROUP BY "Mission_Outcome"

33
Boosters Carried Maximum Payload

• Following boosters carried the maximum payload of

15600 kg SQL query is as below.
1. F9 B5 B1048.4
%sql SELECT DISTINCT "Booster_Version" FROM
2. F9 B5 B1049.4
SPACEXTBL WHERE "PAYLOAD_MASS__KG_" IS ( SELECT
3. F9 B5 B1051.3
MAX("PAYLOAD_MASS__KG_") FROM SPACEXTBL)
4. F9 B5 B1056.4

5. F9 B5 B1048.5

6. F9 B5 B1051.4

7. F9 B5 B1049.5

8. F9 B5 B1060.2

9. F9 B5 B1058.3

10. F9 B5 B1051.6

11. F9 B5 B1060.3

12. F9 B5 B1049.7

34
2015 Launch Records

• List the failed landing_outcomes in drone ship, their booster versions,

and launch site names for in year 2015
Month Year Landing_Outcome Booster_Version Launch_Site
01 2015 Failure (drone ship) F9 v1.1 B1012 CCAFS LC-40
04 2015 Failure (drone ship) F9 v1.1 B1015 CCAFS LC-40

• SQL Query is as below

• %sql SELECT substr("Date",6,2) as "Month" , substr("Date",0,5) as Year ,
"Landing_Outcome", "Booster_version", "Launch_site" FROM SPACEXTBL WHERE "Landing_Outcome"
IS "Failure (drone ship)" AND SUBSTR("Date",0,5) IS "2015"

35
Rank Landing Outcomes Between 2010-06-04 and 2017-03-20

• Ranked count of landing outcomes.

• SQL query is as below.

• %sql SELECT Landing_Outcome, COUNT(*) as "Events" FROM
SPACEXTBL WHERE "Date" BETWEEN "2010-06-04" AND "2017-
03-20" GROUP BY "Landing_Outcome" ORDER BY "Events"
DESC

• We are selecting the count of landing outcomes

between the given dates by grouping them and
ordering them in descending order.

36
Section 3
SpaceX Launch Sites
Key considerations selecting a space
launch sites are as below, which can be
observed from here and other assessments
below.
1. Proximity to the Equator:
Earth's Rotational Velocity provides an initial boost to a rocket
launched eastward. This boost is maximized closer to the
equator, where the Earth's rotational speed is highest. And there
is also Orbital Inclination which Launching closer to the equator
allows for easier access to a wider range of orbital inclinations,
including geostationary orbits.
2. Downrange Safety:
Special considerations such as using unpopulated areas where
Launch sites are typically located in remote areas with minimal
population density to minimize the risk of casualties in case of
launch failures. Further Many launch sites are situated near
large bodies of water to ensure that falling debris lands in a safe,
unpopulated area.
3. Infrastructure and Accessibility:
Transportation: Good transportation links are essential for
transporting rocket components, fuel, and personnel to the
launch site.
Support Facilities: Adequate infrastructure, including power,
water, and communication systems, is necessary for the
operation of the launch site.

38
Launch Outcomes

• Only 7/26 ( 27%) of the

launches made from CCAFS LC -
40 were sucessfull

• Only 3/7 (43%) of the launches

were successful form CCAFS
SLC-40

Green for Successful Launches 39

Red for Unsuccessful launches

Launch Outcomes

• Only 4/10 (40%) of the launches made

from VFAB SLC-4E were successful.

However
• 10/13 ( 77%) of the launches made
from KSC-LC 39A were successful

Green for Successful Launches 40

Red for Unsuccessful launches

Distance to Proximities from site VFAB SLC-4E

• VFAB SLC-4E site is 0.89 km

from the sea

• And it is 16.5 km from the

nearest city Titusville

• Close to main transportation

locations such as 5.6 km from
the railway and .84 km form the
highway
41
Section 4
All Site Launch Success

• KSC LC-39A has the

highest number of
successful launches.

43
KSC LC-39A Site Success Rate

• There are 10(77%)

Successfully launches
from KSC LC-39A site

44
Impact on Payload Mass to Launch Success

• There seems to be limited success to

launches beyond 6000 kg

• Also not much of a success for

payloads lower than 2000 kg

• Hence data indicate the optimum

payload range to be between 2000
and 6000
45
Section 5
Classification Accuracy

• Based on the accuracy factor, Decision

Tree Classifier has the highest accuracy
followed by Logistic Regression

47
Confusion Matrix of Decision Tree Classifier
• A confusion matrix summarizes the performance of a
classification algorithm
• The fact that there are false positives (Type 1 error) is not a good
indicator, which reduces the precision and F1 score
• Confusion Matrix Outputs are as below for the decision tree
classifire:
• 12 True positive
• 2 True negative
• 4 False positive
• 0 False Negative

• Precision= TP / (TP + FP) = 12 / 16 = .75

• Recall= TP / (TP + FN)12 / 12 = 1

• F1 Score= 2 * (Precision * Recall) / (Precision + Recall)

=2 * (.75 * 1) / (.75 + 1) = 1,5/ 1.75 = 0.857

• Accuracy= (TP + TN) / (TP + TN + FP + FN) = 14/18

= 0.77

48
Conclusions

• Success of missions has increase with time

• We can observe a 100% success rate for ES-L1, GEO , HEO and SSO orbit types
• There were serval drone ship landings
• Highest performing payload mass is between 2,000 and 6,000 kg
• Space X has chosen sites near the equator. Other factors such as close to a main road and railway was
also import
• Being close to a coast makes some room for failed landings
• Most successful site is KSC LC-39A with a success rate of 77%
• Decision Tree Classifier has the best learning algorithm for the data available

49