Once I found the Top 50 features, I scaled everything down to achieve a higher accuracy with values being more alike. After scaling the data to a standard format, I conducted a test-train split. 75% of the data was used for the training data and the other remaining 25% become the test data.
I then performed a logistic regression and SVC to compare results. A classification report was used on each to verify results. The machine yielded adequate results between 96% and 97% in the slight favor of SVC. This suggests that it can predict the likelihood of a company going sinking to the bottom of the financial sea with more accuracy than logistic regression.
After analyzing multiple rows of data from WHO over the past 10 years, I will describe what I found.
First, I will start by saying that the data I got contained several csv files of living standard metrics worldwide. Examples include birth rates, suicide rates, cancer rates, abuse rates, etc. I performed EDA (exploratory data analysis) after purging the data of unnecessary columns, null values, and other inconsistencies that would lead to inaccurate results from analysis. The quality of the data is in question for me. The amount of inconsistencies compromised the integrity of the overall data leaving little room for further analysis.