This paper discusses the practical issues of working with data sets common to finance, statistics, and other related fields, focusing on the Python library pandas. Pandas is designed to facilitate working with these data sets and provide fundamental building blocks for statistical models. The paper discusses design issues encountered in developing pandas, compares it with R, and outlines future directions for statistical computing and data analysis using Python. Python is increasingly used in scientific applications traditionally dominated by R, MATLAB, Stata, SAS, etc. The maturity of numerical libraries like NumPy and SciPy, along with good documentation and available distributions like EPD and Pythonxy, have made Python accessible and convenient for a broad audience. However, adoption of Python for applied statistical modeling has been slower compared to other areas of computational science. A major issue for statistical Python programmers has been the lack of libraries implementing standard models and a cohesive framework for specifying models. However, recent developments in econometrics, Bayesian statistics, and machine learning have improved the situation. Despite this, many statisticians still prefer R due to its domain-specific nature and well-vetted open-source libraries. The paper focuses on data structures and tools for working with data sets in-memory, which are fundamental for constructing statistical models. Pandas is a new Python library of data structures and statistical tools, initially developed for quantitative finance applications. It provides a DataFrame class similar to R's data.frame, with enhancements like built-in data alignment. Pandas provides a DataFrame class that implements much of the functionality of its R counterpart, with some important enhancements. It can handle structured data, reshape it, and perform operations like data alignment and missing data handling. Pandas also supports panel data (3D data sets) and provides tools for combining or joining data sets. The paper discusses the use of NaN to represent missing data, the handling of categorical variables and "group by" operations, and the implementation of statistical models. Pandas also provides efficient date and time handling, which is crucial for time series data. The paper concludes that pandas represents a solid step in the right direction for making Python a compelling choice for data analysis applications. It also mentions related packages and future development work, including collaboration with scikits.statsmodels to improve statistical modeling tools in Python.This paper discusses the practical issues of working with data sets common to finance, statistics, and other related fields, focusing on the Python library pandas. Pandas is designed to facilitate working with these data sets and provide fundamental building blocks for statistical models. The paper discusses design issues encountered in developing pandas, compares it with R, and outlines future directions for statistical computing and data analysis using Python. Python is increasingly used in scientific applications traditionally dominated by R, MATLAB, Stata, SAS, etc. The maturity of numerical libraries like NumPy and SciPy, along with good documentation and available distributions like EPD and Pythonxy, have made Python accessible and convenient for a broad audience. However, adoption of Python for applied statistical modeling has been slower compared to other areas of computational science. A major issue for statistical Python programmers has been the lack of libraries implementing standard models and a cohesive framework for specifying models. However, recent developments in econometrics, Bayesian statistics, and machine learning have improved the situation. Despite this, many statisticians still prefer R due to its domain-specific nature and well-vetted open-source libraries. The paper focuses on data structures and tools for working with data sets in-memory, which are fundamental for constructing statistical models. Pandas is a new Python library of data structures and statistical tools, initially developed for quantitative finance applications. It provides a DataFrame class similar to R's data.frame, with enhancements like built-in data alignment. Pandas provides a DataFrame class that implements much of the functionality of its R counterpart, with some important enhancements. It can handle structured data, reshape it, and perform operations like data alignment and missing data handling. Pandas also supports panel data (3D data sets) and provides tools for combining or joining data sets. The paper discusses the use of NaN to represent missing data, the handling of categorical variables and "group by" operations, and the implementation of statistical models. Pandas also provides efficient date and time handling, which is crucial for time series data. The paper concludes that pandas represents a solid step in the right direction for making Python a compelling choice for data analysis applications. It also mentions related packages and future development work, including collaboration with scikits.statsmodels to improve statistical modeling tools in Python.