Cloudera CDP Data Analyst study plan

Write and run ten Hive queries against a sample table that exercise SELECT, WHERE, GROUP BY, HAVING, ORDER BY, and LIMIT, then annotate each with the execution order of its clauses.

Distinguish Hive from Impala by listing at least six behavioral differences (execution engine, latency, DDL semantics, transactions, type coercion, result caching) and verifying three of them with live queries.

Describe the CDP data analyst role by mapping where Hive, Impala, Kudu, Spark, and BI tools sit in the platform stack, and state which engine you would pick for three given workloads.

Practice reading a query plan with EXPLAIN in Impala and identify the scan, join, and exchange stages in three plans.

Construct INNER, LEFT, RIGHT, and FULL OUTER joins across three related tables and explain how NULLs propagate into aggregates in each case.

Write queries using window functions (ROW_NUMBER, RANK, LAG/LEAD, SUM OVER) to solve a running-total and a top-N-per-group problem.

Implement semi-join and anti-join patterns in both Hive and Impala and compare their syntax and performance characteristics.

Apply date and string functions to bucket timestamps and parse delimited fields, and document the return type of each function used.

Design and create internal, external, and partitioned Hive tables, then load data and confirm partition pruning appears in EXPLAIN output.

Profile a raw dataset for nulls, duplicates, outliers, and inconsistent formats using aggregate queries, and record findings in a data-quality log.

Transform the profiled dataset into a cleaned analytical table using CASE, COALESCE, CAST, and deduplication logic.

Choose appropriate file formats and compression (Parquet, ORC, Avro, Snappy, Zstd) for three given access patterns and justify each choice.

Create a Kudu table with a well-chosen primary key and range partitioning, then demonstrate the effect of key order on a range-scan query.

Perform upserts, deletes, and updates on Kudu data from Impala and verify row counts after each operation.

Explain when to choose Kudu versus HBase versus HDFS-backed Parquet for three scenarios involving random access, scan throughput, and update frequency.

Write an Impala query joining a Kudu dimension table to a Parquet fact table and describe how the join strategy is selected.

Read and write the same dataset with Spark DataFrame API and Spark SQL, then confirm both produce identical results.

Apply transformations and actions to build an aggregation pipeline, and inspect the physical plan to identify a shuffle.

Use Cloudera Data Engineering to run a submitted Spark job on a schedule and retrieve its logs and status.

Convert a Hive query from week 2 into a Spark job and note where syntax or semantics had to change.

Connect a BI tool to Impala or Hive through the appropriate driver, and configure a live connection versus an extract.

Build a dashboard with at least four visual types and document the grain, filters, and aggregation behind each visual.

Create Ranger policies restricting a column and a row set for a test user, then verify the restriction by querying as that user.

Trace lineage for an analytical table in Atlas and identify its upstream source tables and downstream consumers.

Diagnose three slow queries using EXPLAIN, runtime profiles, and table statistics, and apply a fix to each (partitioning, join order, statistics refresh, or file sizing).

Collect and refresh table and column statistics in Hive and Impala, and demonstrate the plan change before and after.

Identify and resolve small-file problems by compacting a partition and measuring query-time improvement.

Explain how Ranger masking and row filters affect query results and how Atlas audit records can be used to investigate access.

Complete a timed 60-question practice set weighted by domain percentage and score each domain separately.

Re-study only the two weakest domains identified by the practice set and rewrite your notes for them.

Rehearse explaining five core concepts aloud (partitioning, Kudu key design, shuffle, Ranger column masking, statistics) in under two minutes each.

Review all prior deliverables and consolidate them into a single indexed reference document.