ElastiCube Python Bindings

High-performance OLAP cube library for Python, powered by Rust, Apache Arrow, and DataFusion.

Features

• Fast: Native Rust implementation with near C-level performance
• Columnar Storage: Efficient memory layout using Apache Arrow
• Familiar API: Pandas-like interface for data scientists
• Type Safe: Full type hints and stub files for IDE support
• No Pre-Aggregation: Query raw data with dynamic aggregations

Installation

pip install elasticube

Quick Start

from elasticube import ElastiCubeBuilder

# Build a cube from CSV data
builder = ElastiCubeBuilder()
builder.add_dimension("region", "string")
builder.add_dimension("date", "date")
builder.add_measure("sales", "float64", "sum")
builder.add_measure("quantity", "int32", "sum")
builder.load_csv("sales_data.csv")

cube = builder.build()

# Query the cube
query = cube.query()
query.select(["region", "SUM(sales)", "SUM(quantity)"])
query.filter("date >= '2024-01-01'")
query.group_by(["region"])
query.order_by(["region"])

# Get results as Pandas DataFrame
df = query.to_pandas()
print(df)

Supported Data Types

• Numeric: int32, int64, float32, float64
• Text: string (utf8)
• Temporal: date, timestamp
• Boolean: bool

Aggregation Functions

• sum - Sum of values
• avg / mean - Average
• min - Minimum value
• max - Maximum value
• count - Count of non-null values
• count_distinct - Count of unique values
• median - Median value
• stddev - Standard deviation
• variance - Variance

Data Sources

ElastiCube supports multiple data source formats:

• CSV: builder.load_csv("data.csv")
• Parquet: builder.load_parquet("data.parquet")
• JSON: builder.load_json("data.json")

License

Dual licensed under MIT or Apache-2.0.