pandas (always written in lower case) is an open-source high-level and high-performance library for data analysis in the Python ecosystem.
Although numpy already offers the ndarray object and many relevant functions for data processing, working with numpy remains at a low performance level as the data representation is most of the time not user friendly.
By contrast, pandas provides easy-to-use labeled data structures and high level data manipulation tools (based on Python dictionaries and numpy arrays) rendering data analysis and processing extremely more understandable and efficient for humans.
Overall, pandas is built on top of numpy and offers to scientist/programmer sophisticated data manipulation features that are expected to significantly increase productivity.
The library name comes from panel data, a term for multidimensional data sets in statistics and econometrics.