Usage

For data manipulation, there are two types of verbs;

1. One table verbs

2. Two table verbs

We define the usage by how the verbs accept the data argument. There are three types of usage;

1. Piping - The data is to the left side of the pipe symbol (>>).

2. Composing - The data is the first argument of the verb.

3. Currying - The data is the only argument of an instantiated verb.

The single table verbs support all three types, while two table verbs only support composing. Here is an example of the same operation in three different ways.

df = pd.DataFrame({'x': [0, 1, 2, 3, 4, 5]})

# Piping
df >> define(w='x%2', y='x+1', z='x+2.5') >> arrange('w')

# Composing
arrange(define(df, w='x%2', y='x+1', z='x+2.5'), 'w')

# Currying
arrange('w')(define(w='x%2', y='x+1', z='x+2.5')(df))

Although composing is the normal way calls are invoked, since data manipulation often involves consecutive function calls, the nested invocations become hard to read. Piping helps improve readability. Currying exists only to spite the zen of python.

Data mutability

By default, plydata does not modify in input dataframe. This means that the verbs have no side effects. The advantage that the user never worries about creating copies to avoid contaminating the input dataframe. It is normal in most data-analysis workflows for the user to manipulate the data many times in different ways. It is also the case that most datasets are small enough that they can be copied many times with no noticeable effect on performance.

If you have a dataset small enough to fit in memory but too large to copy all the time without affecting performance, then consider using the modify_input_data option. However, only a few verbs can modify the input data and it is noted in the documentation.

Datastore Support

Single Table Verbs

Verb	Dataframe	Sqlite
arrange	Yes	No
create	Yes	No
define	Yes	No
distinct	Yes	No
do	Yes	No
group_by	Yes	No
group_indices	Yes	No
head	Yes	No
pull	Yes	No
query	Yes	No
rename	Yes	No
sample_frac	Yes	No
sample_n	Yes	No
select	Yes	No
slice_rows	Yes	No
summarize	Yes	No
tail	Yes	No
ungroup	Yes	No

Helper verbs

Verb	Dataframe	Sqlite
count	Yes	No
tally	Yes	No
add_count	Yes	No
add_tally	Yes	No
call	Yes	No
arrange_all, arrange_at, arrange_if	Yes	No
create_all, create_at, create_if	Yes	No
group_by_all, group_by_at, group_by_if	Yes	No
mutate_all, mutate_at, mutate_if	Yes	No
query_all, query_at, query_if	Yes	No
rename_all, rename_at, rename_if	Yes	No
summarize_all, summarize_at, summarize_if	Yes	No

Two table verbs

Verb	Dataframe	Sqlite
anti_join	Yes	No
inner_join	Yes	No
left_join	Yes	No
outer_join	Yes	No
right_join	Yes	No
semi_join	Yes	No