MaxFrame APIs

MaxFrame provides two API categories: compatibility APIs that integrate with standard libraries such as Pandas for data processing, and MaxFrame-specific APIs for distributed task execution. Using MaxFrame APIs lets you obtain data manipulation experience similar to that of standard databases and efficiently run large-scale data processing tasks on the MaxFrame platform.

MaxFrame-specific APIs

API type	API name	Description
Task session	new_session	Initializes a MaxFrame job session. Call new_session at the beginning of your script to create a session object that interacts with the backend service for subsequent data processing.
Input/Output	read_odps_table	Builds a DataFrame from a MaxCompute table, or from specific partitions or columns of the table.
	read_odps_query	Builds a DataFrame from MaxCompute SQL query results.
	to_odps_table	Writes DataFrame data to a MaxCompute table.
Task execution	Execute	Triggers task execution. MaxFrame uses lazy evaluation: operations are deferred until execute is called, allowing data conversions to be delayed to the end of the pipeline so that multiple operations can be executed at the same time in a single task. This reduces client-server data transfer and improves performance.
Task result retrieval	Fetch	Retrieves task results to the client. The computing result is not directly transmitted to the client, which reduces data transfer and client workload. Call fetch after execute to pull specific result data on demand.

MaxFrame for Pandas

DataFrame

API type	Reference
Constructor	DataFrame
Attributes	Attributes and underlying data
Math operations	Binary operator functions Computations / descriptive stats
Filtering, projection, and sampling	Reindexing / selection / label manipulation
Sorting	Reshaping / sorting / transposing
Combining, joining, and merging	Combining / joining / merging
User-defined functions (UDFs), aggregates, and windows	Function application / groupby / window
Conversion	Conversion

Series

API type	Reference
Constructor	Series
Math operations	Binary operator functions Computations / descriptive stats
Filtering, projection, and sampling	Reindexing / selection / label manipulation
Sorting	Reshaping / sorting
UDFs, aggregates, and windows	Function application / groupby / window

Index

API type	Reference
Constructor	Index
Filtering, projection, and sampling	Reindexing / selection / label manipulation
Conversion	Conversion

Pandas APIs supported by MaxFrame

DataFrame
- APIs that support all parameters:
  
  DataFrame, add, radd, round, sub, rsub, mul, rmul, floordiv, rfloordiv, truediv, rtruediv, div, rdiv, mod, rmod, pow, rpow, eq, ne, lt, gt, le, head, reset_index, drop_duplicates, rename, sort_values, sort_index, join, merge, sample, transform, and agg
- APIs that support specific parameters:
  - astype: does not support astype('category').
  - set_axis: supports only axis=1.
  - apply: supports only axis=1.
Series
- APIs that support all parameters:
  
  add, radd, round, sub, rsub, mul, rmul, floordiv, rfloordiv, truediv, rtruediv, div, rdiv, mod, rmod, pow, rpow, eq, ne, lt, gt, ge, le, head, drop, reset_index, drop_duplicates, sort_values, sort_index, sample, transform, and agg
- APIs that support specific parameters: None
Index
- APIs that support all parameters: drop_duplicates
- APIs that support specific parameters:
  
  astype: does not support astype('category').