Your PySpark playground
Explore DataFrames, write transformations and practice SQL. No installation. Sign in to run code.
—Learn by doing
Choose a small, self-contained example. It opens in a new cell, so your existing code stays in place. Sign in and select Run to see the result.
Create a small dataset and keep the rows you need with a filter.
people = spark.createDataFrame(
[("Ada", 32), ("Lin", 24)],
["name", "age"]
)
people.filter("age >= 30").show()Then Run: One row: Ada, age 32.
The filter keeps only rows where age is at least 30. Lin is excluded.
Next exercise: Change the condition to age < 30. Predict the row, then run it.
Related challenge: Filter Rows with a ConditionCalculate totals by category with the PySpark DataFrame API.
from pyspark.sql import functions as F
sales = spark.createDataFrame(
[("Books", 20), ("Books", 15)],
["category", "amount"]
)
sales.groupBy("category").agg(
F.sum("amount").alias("total")
).show()Then Run: One row: Books, total 35.
groupBy combines the two Books rows; sum adds 20 and 15.
Next exercise: Add a Toys row with amount 10. Predict the two grouped totals.
Related challenge: Count Rows per CityRegister a DataFrame as a view, then query it inside a PySpark cell.
orders = spark.createDataFrame(
[(1, 25), (2, 80)], ["id", "amount"]
)
orders.createOrReplaceTempView("orders")
spark.sql(
"SELECT * FROM orders WHERE amount > 50"
).show()Then Run: One row: id 2, amount 80.
The temporary view makes the DataFrame available to Spark SQL. WHERE filters its rows.
Next exercise: Change the query to SELECT SUM(amount) AS total FROM orders. Expect 105.
Related challenge: Select Two Columns from a DataFrameSave explicitly in Notebook tools to recover code after refresh. Outputs and server variables are temporary. Execution and resource limits are listed in Notebook tools.
Start with the example, sign in, then select Run or press Shift+Enter inside the editor. Add cells for each transformation; Run All executes in order and stops at the first error.
Open Data to upload and preview datasets or inspect session DataFrames. Python, Pandas, Matplotlib and NLP cells share the Python runtime. SQL cells use a separate SQLite database; use spark.sql() in a PySpark cell for Spark SQL.
What to expect from your notebook, SQL cells and execution session.
Yes. You can explore the editor and examples before signing in. Sign in with Google to execute code and upload practice data. No local Spark installation is needed.
SQL cells run in a separate SQLite database. To query a Spark DataFrame, create a temporary view and call spark.sql() inside a PySpark cell. SQLite tables and Spark views are separate.
Use Notebook tools to explicitly Save on this device, then Recover after a refresh or signing back into the same account. Later edits need another Save. Device drafts do not sync automatically; export/import a private notebook JSON file to move code between devices. Outputs and server variables are not restored.
Stop skips the remaining cells in a batch after the current cell finishes. It does not immediately cancel a running backend operation. Run requests wait up to 150 seconds; use small, bounded examples for practice.
Yes. This notebook lets you run PySpark in the browser, so you can test code, explore DataFrames, and practice transformations without a local Spark setup.
Yes. Add cells to ingest sample or uploaded data, clean it, join tables and build aggregations in one notebook. Python and PySpark cells share variables; SQLite has its own tables.
You can work with PySpark, Python, SQL, Pandas, Matplotlib, and NLP cells, which makes the page useful for end-to-end data engineering experiments.
It is a strong fit for beginners learning Spark, data engineers prototyping ETL logic, and interview candidates who want a fast PySpark playground.