SequenceExpr represents a column in a two-dimensional dataset. You cannot manually create a SequenceExpr object. Instead, you can retrieve one from a Collection object.

Column retrieval

You can use collection.column_name to retrieve a column. 
iris.sepallength.head(5)
   sepallength
0          5.1
1          4.9
2          4.7
3          4.6
4          5.0
If the column name is stored in a string variable, you can use df[column_name] to retrieve a column. 
iris['sepallength'].head(5)
   sepallength
0          5.1
1          4.9
2          4.7
3          4.6
4          5.0

Column types

DataFrame has its own type system. When a table is initialized, the MaxCompute data types are cast. This design provides support for more types of execution backends. The execution backend of DataFrame can be MaxCompute SQL, Pandas, or a MySQL or PostgreSQL database.

The following table lists the mappings between MaxCompute and DataFrame data types.
MaxCompute type DataFrame type
BIGINT INT64
DOUBLE FLOAT64
STRING STRING
DATETIME DATETIME
BOOLEAN BOOLEAN
DECIMAL DECIMAL
ARRAY<VALUE_TYPE> LIST<VALUE_TYPE>
MAP<KEY_TYPE, VALUE_TYPE> DICT<KEY_TYPE, VALUE_TYPE>
When options.sql.use_odps2_extension=True, the following data types are also supported.
TINYINT INT8
SMALLINT INT16
INT INT32
FLOAT FLOAT32
The following items describe DataFrame data types:
  • PyODPS DataFrame supports the following data types: INT8, INT16, INT32, INT64, FLOAT32, FLOAT64, BOOLEAN, STRING, DECIMAL, DATETIME, LIST, and DICT.
  • For LIST and DICT types, you must specify the types of the values they contain. Otherwise, an error occurs.
  • DataFrame does not support the TIMESTAMP and STRUCT types introduced in MaxCompute 2.0. The new types will be supported in future releases.
  • You can use sequence.dtype to retrieve the data type of a Sequence object. 
    iris.sepallength.dtype
float64
  • You can use the astype method to change the type of a column. This method requires a type as input and returns the converted Sequence object as output. 
    iris.sepallength.astype('int')
   sepallength
0            5
1            4
2            4
3            4
4            5

Column name

In DataFrame computing, a Sequence object must have a column name. Typically, DataFrame automatically creates a name for each Sequence object.
iris.groupby('name').sepalwidth.max()
   sepalwidth_max
0             4.4
1             3.4
2             3.8

In the preceding example, sepalwidth is named sepalwidth_max after the max() method is applied to take the maximum value. In some operations, such as adding a Scalar object to a Sequence object, the resulting column is automatically assigned the name of the Sequence object. In other cases, you need to manually name a Sequence object.

You can use the rename method to rename a Sequence object. 
iris.sepalwidth.rename('sepal_width').head(5)
   sepal_width
0          3.5
1          3.0
2          3.2
3          3.1
4          3.6

Simple column transformations

You can perform operations on a Sequence object to obtain a new Sequence object. This is similar to performing operations on a simple Python variable.
(iris.sepallength + 5).head(5)
   sepallength
0         10.1
1          9.9
2          9.7
3          9.6
4         10.0
Note that when two columns are involved in operations, you need to manually name the resulting column. For more information about column transformations, see Column operations. 
(iris.sepallength + iris.sepalwidth).rename('sum_sepal').head(5)
   sum_sepal
0        8.6
1        7.9
2        7.9
3        7.7
4        8.6