Introduction:
In the vast expanse of the cosmos, human curiosity knows no bounds. Since the inception of space exploration with the launch of the USSR's Sputnik, our journey into the unknown has been fueled by the relentless pursuit of knowledge and technological advancement. In this article, we delve into a Data Science project centered around Space Mission Exploratory Data Analysis (EDA) and predictive modeling.
Understanding the Dataset:
The dataset at hand encapsulates a treasure trove of information, ranging from launch dates and locations to the success rates of space missions. It provides a comprehensive view of the endeavors undertaken by various space agencies and private companies. To kickstart the analysis, we focused on key aspects such as DateTime, Year, and the geographical location specified by city and country. This groundwork laid the foundation for a deeper exploration.
Top Companies and Launch Frequency:
One intriguing facet of the analysis involved identifying the top 15 companies involved in satellite missions. By leveraging the dataset, we shed light on the frequency of launches over the years. This information not only provides insights into the dynamic nature of space exploration but also highlights the key players in this ambitious field.
Analyzing ISRO's Success Rate:
A closer look at the dataset allowed us to focus on the Indian Space Research Organisation (ISRO). Through meticulous analysis, we gauged ISRO's success rates and visualized year-wise failed missions. This exploration not only serves as a testament to ISRO's dedication but also provides valuable insights for future endeavors.
Feature Engineering and Prediction:
With a solid understanding of the dataset, we ventured into feature engineering and prediction. Unsurprisingly, challenges emerged, including missing data on 3360 rocket launches. A judicious decision to drop unnecessary features like location and datetime streamlined the dataset, setting the stage for predictive modeling.
Addressing Missing Data:
Handling missing data is a critical aspect of any data science project. Leveraging the SimpleImputer function from the sklearn.impute module, we tackled the NaN values in our dataset. This strategic move ensured the robustness of our analysis and paved the way for accurate predictive modeling.
Predictive Modeling – Logistic Regression and RandomF orest Classifier:
Two formidable models, Logistic Regression and Random Forest Classifier, were employed for prediction. Despite the latter outperforming the former, the Random Forest model exhibited suboptimal recall and F1 scores, indicating its limitations. This revelation prompts a deeper exploration of the data and potential avenues for model improvement.
Conclusion:
Embarking on a data-driven journey through space exploration showcases the intersection of technology, science, and human curiosity. Our EDA and predictive modeling project offers a glimpse into the intricacies of space missions, providing a foundation for future analyses and advancements. As we navigate the cosmos, one dataset at a time, the pursuit of knowledge propels us towards the final frontier.
Top comments (0)