SFR
Analysis
Project Overview
The Space Fund Realty (SFR) Analysis project aims to provide valuable insights into aerospace companies and their missions, ultimately assisting investors in making informed decisions. The SFR is a crucial rating system that evaluates companies based on their missions, payload, launch costs, and other factors, providing an indication of their development and stability. The SFR rating ranges from 1 to 9, with higher ratings signifying more developed companies.
Data Dictionary
| Column Name | Description |
|---|---|
| Company | Name of the company |
| SFR | SpaceFund Realty rating of the company |
| Payload(kg) | Payload of the mission |
| Launch Cost(million USD) | Launch cost of the mission |
| Price per kg | Price per kg payload of the mission |
| Launch Class | Launch class of the mission |
| Orbit Altitude | Orbit altitude of the mission |
| Tech Type | Technology type of the mission |
| Country | Country of the company |
| HQ Location | Headquarters location of the company |
| Description | Description of the mission |
Conclusion
From the exploratory data analysis, I have conclded that most of the companies in the dataset were from US which resulted in greater comapnies with good SFR from US. However, in case of China, the number of companies is less than UK, but still it ranks secound with number of companies with SFR greater than 6. Majority of the missions were rocket type, small launch class and Low Earth Orbit missions. This relation has been found due to similar SFR distribution in their resspective graphs, where most of the missions have 2-3 SFR.
There has been a similar relationship of launch cost and payload with the SFR. Missions with higher launch cost and higher payload have higher SFR. This shows that the companies which are well established will be able to take more payload to space and spend more money on their missions. The companies with poor SFR are in their initial phase of development and are not capable of taking heavy payloads to space and spending more money on their missions.
Coming to the machine learning models, I have used Decision Tree and Random Forest Classifier. Both the models have given similar result with similar accuracy, i.e. 87%. However, due to small dataset, the models had very poor recall score in predictiong the SFR greater than 6. This could be improved by increasing the dataset.
View Notebook Download Dataset