An overview of the analysis:
Purpose of this study is not analyse the credit risk of potential lenders. Using a dataset of historical lending activity from a peer-to-peer lending services company to build a model that can identify the creditworthiness of borrowers.
The results: Using a bulleted list, describe the accuracy score, the precision score, and recall score of the machine learning model.
- Accuracy is the ratio of correctly predicted observations to the total observations. It measures how often the model is correct overall. The model predicted 99% of all instances in the overall dataset.
- Precision is the ratio of correctly predicted positive observations to the total predicted positives. Precision answers the question: "What proportion of positive identifications was actually correct?", the weighted average of this for the model for this amounted to 99% precision, signifying high level reliance.
- Recall is the ratio of correctly predicted positive observations to all observations in the actual class. Recall answers the question: "What proportion of actual positives was identified correctly?". This was also very successful showcasing 99% efficiency on the weighted average across the entire dataset.
A summary: Summarize the results from the machine learning model. Include your justification for recommending the model for use by the company. If you don’t recommend the model, justify your reasoning.
These results suggest that the model is highly accurate, especially in predicting the healthy loans, but has some potential for improvement in predicting 1 (the high risk loans) . This is common in scenarios where there is a class imbalance, with the model performing better on the majority class / 0 (healthy loans).
With an accuracy rate of 99%; pointing to a highly effective model and even with the slight discrepancy in performance between the two sets of loan groups the weighted average still points to being highly accurate across the total population sample.