This paper discusses the practical issues of working with data sets common in finance, statistics, and related fields, focusing on the pandas library. Pandas is designed to facilitate data manipulation and provide a set of fundamental building blocks for implementing statistical models. The paper highlights the design challenges encountered during the development of pandas and compares it with R, emphasizing its strengths and potential for statistical computing. Key features of pandas include: 1. **Data Structures**: pandas introduces DataFrame and Series objects, which are similar to R's data.frame and data.table, but with additional enhancements like built-in data alignment. 2. **Handling Missing Data**: pandas uses NaN to represent missing data, which is more efficient than NumPy MaskedArrays but has limitations in certain contexts. 3. **Combining Data Sets**: pandas supports operations like merging, joining, and grouping, which are crucial for data manipulation. 4. **Panel Data**: pandas provides classes for handling 3-dimensional data, such as LongPanel and WidePanel, which are useful for econometric applications. 5. **Statistical Models**: pandas can be used to implement statistical models, including ordinary least squares regression, with minimal data preparation. 6. **Date/Time Handling**: pandas offers efficient tools for working with date and time data, leveraging Python's built-in datetime type. The paper concludes by discussing future directions for statistical computing and data analysis using Python, emphasizing the potential for pandas to become a compelling choice for data analysis applications.This paper discusses the practical issues of working with data sets common in finance, statistics, and related fields, focusing on the pandas library. Pandas is designed to facilitate data manipulation and provide a set of fundamental building blocks for implementing statistical models. The paper highlights the design challenges encountered during the development of pandas and compares it with R, emphasizing its strengths and potential for statistical computing. Key features of pandas include: 1. **Data Structures**: pandas introduces DataFrame and Series objects, which are similar to R's data.frame and data.table, but with additional enhancements like built-in data alignment. 2. **Handling Missing Data**: pandas uses NaN to represent missing data, which is more efficient than NumPy MaskedArrays but has limitations in certain contexts. 3. **Combining Data Sets**: pandas supports operations like merging, joining, and grouping, which are crucial for data manipulation. 4. **Panel Data**: pandas provides classes for handling 3-dimensional data, such as LongPanel and WidePanel, which are useful for econometric applications. 5. **Statistical Models**: pandas can be used to implement statistical models, including ordinary least squares regression, with minimal data preparation. 6. **Date/Time Handling**: pandas offers efficient tools for working with date and time data, leveraging Python's built-in datetime type. The paper concludes by discussing future directions for statistical computing and data analysis using Python, emphasizing the potential for pandas to become a compelling choice for data analysis applications.