v0.6.8

• DataFrame addfillna and dropna for collection
• Bug fixes

v0.6.7

• DataFrame add more aggregation including moment, skew, kurtosis, and cat.
• DataFrame reshuffle API
• DataFrame optimize MapReduce performance, recommend upgrade if use map_reduce
• Fix segment fault which may happen when downloading data through cython
• Refactor and bugfixes

v0.6.6

• odps.ml is available for machine learning
• DataFrame add extract_kv, to_kv, melt API
• DataFrame add preprocess API including min_max_scale, std_scale, split, append_id
• DataFrame.describe()'s result is more readable
• DataFrame.concat on axis=0
• DataFrame's user defined function including map, apply and so on support partital function
• Bug fixes

v0.6.5

• Mod support for DataFrame sequence
• Bug fixes

v0.6.4

• DataFrame pivot and pivot_table API
• Fix tunnel bug when uploading data in which \x00 exists with cython optimized code
• Fix DataFrame function codegen which the old one may cause the parameter check failure
• Bugfixes

v0.6.3

• Hotfix performance of tunnel download

v0.6.2

• Fix tunnel, use stream to read
• Bugfixes

v0.6.1

• For DataFrame, astype int64 when encountering a bigint partition field
• For DataFrame, partitioned table now support using tunnel in more cases
• Full Python3 user defined function support

v0.6.0

• DataFrame map, apply, agg and map_reduce now support third-party pure python library
• Paginated DataFrame results and Interactive charts
• Combiner support for MapReduce API
• Extended Sample API
• Auto completer for get_(odps model, table etc)
• Column pruning and predicate pushdown optimization for DataFrame
• Full Python 2.6 support
• Refactoring and bug fixes

v0.5.8

• Add RequestId to ODPSError
• Fix DataFrame bug when scalar is a Chinese StringScalar for ODPS SQL backend
• Fix DataFrame bug when persist a dataframe which has cached dataframe inputs
• Fix DataFrame bug when apply on a projected collection for pandas backend