chDB 4.0 - A pandas Drop-in Replacement Powered by ClickHouse

chDB 4.0 ships a `chdb.datastore` module that acts as a drop-in replacement for pandas — same API, but queries run through ClickHouse under the hood. In this video we explore what that looks like in practice using an Amazon product reviews dataset. What's covered: - Loading a local parquet file with `chdb.datastore.read_parquet()` instead of `pandas.read_parquet()` - Running the same filter + groupby + sort pipeline on both pandas and chDB DataFrames - Using `.explain()` to inspect the lazy query plan chDB builds before execution - Chaining intermediate results (lazy evaluation lets you save and reuse filtered subsets) - Benchmarking pandas vs chDB with `%%timeit` on the same aggregation - Reading a parquet file directly from S3 (s3://datasets-documentation/...) without downloading it first Links: - chDB GitHub: https://github.com/chdb-io/chdb - Amazon reviews dataset: s3://datasets-documentation/amazon_reviews/ - chDB docs: https://clickhouse.com/docs/chdb