This project involves analyzing the effectiveness of different drug regimens on the treatment of squamous cell carcinoma, a commonly occuring form of skin cancer, in mice. Multiple datasets were cleaned, combined, and analyzed using statistical methods and data visualization techniques.
The following tools and libraries were used in this analysis:
- Pandas: for data manipulation and cleaning
- Matplotlib: for data visualization
- Scipy.stats: for statistical analysis
- Numpy: for numerical computing
The study data files, Mouse_metadata.csv and Study_results.csv, were merged into a single dataset. Duplicate mouse data were removed, with one mouse (ID "g989") found to have duplicate data.
Once the data was cleaned, summary statistics were calculated to analyze the tumor volumes for each drug regimen. The results are shown in the following table:
Several different visualizations were created to provide insights into the data. These visualizations include:
- Sample Size: This visualization shows the sample size for each drug regimen used in the study.
- Gender Distribution: This visualization shows the gender distribution of the mice used in the study.
- Final Tumor Volume: This visualization compares the final tumor volume for the top four most effective drug regimens.
- Relationship Between Tumor Volume and Mouse Weight: This visualization shows the relationship between tumor volume and mouse weight as well as the linear regression and correlation coefficient.
Capomulin and Ramicane appear to be the most effective drugs in treating tumors in mice, as they had the lowest mean and median tumor volumes by at least 10 mm3. These drugs also had the lowest variance, standard deviation, and standard error of the mean when looking at tumor volume. Therefore, these results indicate that there is less variability in the effectiveness of these drugs, providing a better idea of what the final tumor volume will be for a mouse that takes them.
There also appears to be a strong positive correlation between the weight of a mouse and the average tumor volume for mice that took Capomulin. Therefore, heavier mice are likely to have larger average tumor volumes.
The data used in this analysis was generated by Mockaroo, LLC using their Realistic Data Generator in 2022.