PyODPS provides a pandas-like interface, PyODPS DataFrame, which operates on MaxCompute tables and can make full use of MaxCompute’s computing power. You can also change the data source from MaxCompute tables to pandas DataFrame, so that the same code can be executed on pandas.
Though PyODPS DataFrame looks like pandas DataFrame, it is not pandas though. Features of pandas, such as full supports for Series and Index, reading data by rows, concatenation of multiple DataFrame by columns are not supported in PyODPS DataFrame. Therefore please check this document before code to make sure methods are supported.