API overview

Import Coral functions from their named module in a Reef project. For example, import Coral.Frame (from_pairs, get_float_col) makes those two names available to your program. Import constructors the same way: FloatCol, IntCol, StringCol, BoolCol, the ColumnType values IntType, FloatType, StringType, BoolType, and the GroupBy specs AggSum, AggMean, AggCount, AggMin, AggMax.

In the signatures below, F[n] stands for Frame[n] and T[n, d] for tensor[n, d]. A result F[k] or F[m] has a row count fixed only when the function runs.

Coral.Frame

Function	Signature	Guide
from_pairs	(pairs: List[(string, Column[n])]) -> F[n]	Construction
from_columns	(cols: Dict[string, Column[n]]) -> F[n]	Construction
empty	(schema: Dict[string, ColumnType]) -> F[n]	Construction
int_col_of_list	(values: List[i64]) -> Column[n]	Construction
get_float_col, get_int_col, get_bool_col	(df: F[n], name: string) -> T[n, f32], T[n, i64], T[n, bool]	Construction
get_string_col	(df: F[n], name: string) -> List[string]	Construction
get_column	(df: F[n], name: string) -> Column[n]	Construction
columns, column_type	(df: F[n]) -> List[string], (df: F[n], name: string) -> ColumnType	Construction
nrows, ncols	(df: F[n]) -> i64	Construction
filter	(df: F[n], mask: T[n, bool]) -> F[k]	Filtering
head, tail	(df: F[n], count: i64) -> F[k]	Filtering
slice	(df: F[n], start: i64, finish: i64) -> F[k]	Filtering
sort_by	(df: F[n], name: string, ascending: bool) -> F[n]	Filtering
with_column, mutate	(df: F[n], name: string, col: Column[n]) -> F[n]	Changing columns
rename	(df: F[n], old_name: string, new_name: string) -> F[n]	Changing columns
drop_column	(df: F[n], name: string) -> F[n]	Changing columns
is_nan, fill_nan, any_nan, count_nan, drop_nan	Float NaN helpers	Missing values
is_nan_col, fill_nan_col, drop_nan_col, any_nan_col, count_nan_col	Integer mask helpers	Missing values
concat	(frames: List[F[n]]) -> F[k]	Concatenation
describe	(df: F[n]) -> F[m]	Describe

Other modules

Module	Functions	Guide
Coral.GroupBy	group_by(df, key_name) -> GroupedFrame[n]; agg_sum, agg_mean, agg_min, agg_max (grouped, col) -> F[m]; agg_count(grouped) -> F[m]; agg(grouped, specs: List[(string, AggFn)]) -> F[m]; value_counts(df, key_name) -> F[m]	GroupBy
Coral.Join	inner_join, left_join, outer_join (left: F[n], right: F[m], on: string) -> F[k]	Joins
Coral.Io	read_csv_frame, read_json_frame (path: string) -> F[n]; write_csv_frame, write_json_frame (df: F[n], path: string) -> unit; read_parquet_frame and write_parquet_frame fail when called	CSV and JSON
Coral.Window	rolling_sum, rolling_mean, rolling_std, rolling_min, rolling_max (col: T[n, f32], window: i64) -> T[n, f32]; ewm(col: T[n, f32], alpha: f32) -> T[n, f32]	Window
Coral.Reshape	melt(df, id_cols, value_cols), pivot(df, index_col, columns_col, values_col), stack(df), unstack(df, index_col), each -> F[m]	Reshape
Coral.AsOf	asof_lookup, asof_join and their _list forms	Sorted-key lookup

See limitations for input and output constraints and for the functions that chelis build rejects.