Spark with Python - Weekend Course - 6 Hours a Day - 24 Hours Course

 Day wise distribution of class :

1. Day 1 : Python, Bigdata , Spark Introduction and Component of Spark ,Operation on Single RDD,
2. Day 2 : Operation on Paired RDD, Fault tolerance and Persistence, Optimizing spark code, IO in spark
3. Day 3 : Spark Streaming, SparkSQL

Detail Deion of class :

Introduction to Big Data and Distributed Computing :

Big data analysis is future. This section of course will help you to understand, the need of distributed computation.

Introduction to data.

Data Science a vision.

• Introduction to Data.
• Big data Introduction.
• Big Data use cases.
• DFS and problem.
• Parallel computation.
• Problem with parallel computation.
• Traditional parallel computation systems

Hadoop :

• Introduction to Hadoop.
• Hadoop Components.
• HDFS and its architecture.
• HDFS Commands
  • mkdir
  • ls
  • rmdir and rm
  • copyFromLocal
  • put
  • cat
  • copyToLocal
  • get
  • touchz
  • mv
  • cp
  • distcp
  • etc…...
• fsimage and edits log files.
• Hadoop property files.
• Introduction to MapReduce.
• Shortcoming of MapReduce.

Python : Refresher

• Introduction to Python.
• Jupytor
• Python variables and Data Type.
• Operators in Python.
• Interactive mode and base programming introduction
• Python Collections (List, Dictionaries etc)
• Control Flow and looping in Python
• Functions in Python (Declaration, Definition Types and calling)
• Object oriented Python.
• NumPy

Spark Introduction :

• Introduction to Spark.
• Spark and Hadoop (Similarity and Differences).
• Spark Execution (Master Slave System , Drive, Driver manager and Executors).
• Spark Shell.
• Resilient Distributed dataSet (RDD).

Operations On RDD :

• Creation of RDD
• Transformation and Action Introduction 
• Lazy evaluation
• Some Important Transformation :
  • Filter
  • map
  • flatMap
  • distinct
  • sample
  • union
  • intersection
  • subtract
  • cartesian
• Some Important Action
  • first
  • take
  • top
  • reduce
  • fold
  • foreach
  • count
  • collect

Paired RDD :

• Introduction and usefulness of Paired RDD.
• Some important Transformation on pairRDD.
  • combineBy
  • mapValues
  • groupByKeys
  • reduceByKeys
  • sortByKeys
  • subsractByKey
  • Joines and their Type
  • cogroup
• Some Important action on pair RDD
  • lookUp
  • collectAsMap
  • countByKey
• Hands on all the functions.

Fault tolerance and Persistence :

• RDD lineage
• persistence
• Benefit of persistence

Optimizing Spark program :

• Introduction to partitioning
• Inbuilt partitioners (Hash and Range)
• Benefits of partitioning
• groupByKey and reduceBykey comparison
• Spark broadcasting and accumulators

IO in Spark :

• TextFile
• Csv File
• JSON
• Data From HDFS

Spark Streaming :

• Introduction to Streaming Data.
• Component of Spark Streaming.
• Transformation 
• Reading from HDFS
• Window Concept
• Push Based Receiver and Pull Based receiver 
• Kafka integration with Streaming.
• State management.
• Performance 

SparkSQL.

• Introduction to SparkSQL
• SparkSQL datatype
• DataFrame an Introduction.
• Creation of a dataframe.
• Summary statistics on DataFrame.
• Aggregation on Given Data.
• SparkSQL and SQL
• Introduction to Hive.
• Using data from Hive and HiveQL.
• Optimizing SparkSQL code.

Spark Code Deployment and cluster managers :

• Submitting Spark code on StandAlone cluster manager.
• Submitting Spark code on YARN
• Submitting Spark code on Mesos

Note : Every part of course will be associated with hands on . A number of objective questions will always help you in scratch your brain.

Projects :

Project 1 : Spark core can be used for data preparation and aggregation. Aggregation will be implemented using Spark core APIs.
For data aggregation movie lance data will be used.

Project 2 : Implementing streaming data word frequency visualization. using Kafka and Spark streaming integration.

Project 3 : Implementation of Moving average using SparkSQL.

Project 4 : Data preprocessing, data manipulation and aggregation using SparkSQL. It will be done using Real time data.