skrub.DataOp.skb.set_data#

DataOp.skb.set_data(data)[source]#

Get a new DataOp with the provided preview values set on the variables.

Returns:

DataOp: A clone of the original DataOp, with the values in data used as the preview values for the corresponding variables.

See also

DataOp.skb.get_data: Obtain the preview values currently set on the variables.
DataOp.skb.get_vars: Obtain the variables (the skrub.var() objects) themselves.
DataOp.skb.clone: Obtain an independent clone of the DataOp, which does not contain any computed preview results. The parameter drop_values controls whether the values set on variables (if any) should be kept.

Examples

>>> import skrub
>>> a = skrub.var('a')
>>> b = skrub.var('b')
>>> c = a + b

We have initialized our variables without any value, there is no preview result available for our DataOp c:

>>> c
<BinOp: add>
>>> c * 2
<BinOp: mul>

As usual, we can evaluate c, passing values for the variables it contains:

>>> c.skb.eval({'a': 1, 'b': 2})
3

But suppose we want to start working in a more interactive fashion and would benefit from preview computations that run whenever we use our DataOp, without needing to explicitly call eval. We can inject values into the existing DataOp.

>>> d = c.skb.set_data({'a': 1, 'b': 2})
>>> d
<BinOp: add>
Result:
―――――――
3
>>> d * 2
<BinOp: mul>
Result:
―――――――
6

(Note that set_data returns a new DataOp (d) and leaves the original one (c) unchanged.)

DataOps that use d can compute previews of the results because unlike c, d has values for its variables:

>>> c.skb.get_data()
{}
>>> d.skb.get_data()
{'a': 1, 'b': 2}

If the DataOp already contained preview values, they are replaced by the provided ones. If we pass values for only some of the variables, any already-existing values for the other variables are kept:

>>> e = d.skb.set_data({'a': 10})
>>> e.skb.get_data()
{'a': 10, 'b': 2}
>>> e
<BinOp: add>
Result:
―――――――
12

If we want to drop the values attached to variables we can use DataOp.skb.clone(). By default it drops values (we can pass drop_values=False to prevent that).

>>> f = e.skb.clone()
>>> f
<BinOp: add>
>>> f.skb.get_data()
{}

When we call set_data, passing keys in data that do not have a corresponding variable in the DataOp is an error.

>>> f.skb.set_data({'x': 0})
Traceback (most recent call last):
    ...
ValueError: The following keys were passed to set_data but have no corresponding variable in the DataOp: ['x']