Why Pig simplifies data transformation in Hadoop - See It in Action

Why Pig Simplifies Data Transformation

📖 Scenario: Imagine you work at a company that collects large amounts of sales data every day. You need to clean and summarize this data to find out which products sell the most. Doing this directly with raw data can be very complex and slow.

🎯 Goal: Learn how to use Apache Pig to simplify data transformation tasks like filtering, grouping, and summarizing big data easily with simple scripts.

📋 What You'll Learn

Create a Pig relation to load sales data

Filter sales records for a specific product category

Group sales by product to calculate total sales

Display the summarized sales results

💡 Why This Matters

🌍 Real World

Companies use Pig to quickly process and analyze large datasets like sales, logs, or user data without writing complex code.

💼 Career

Knowing Pig helps data engineers and analysts handle big data transformations efficiently in Hadoop environments.

1

Load the sales data

Write a Pig Latin statement to load the sales data from the file 'sales_data.csv' into a relation called sales. Assume the data has three fields: product (chararray), category (chararray), and amount (int). Use LOAD with PigStorage(',') and define the schema.

Hadoop

-- Load sales data from 'sales_data.csv' with schema (product:chararray, category:chararray, amount:int)
-- Your code here

Hint

Use LOAD with PigStorage and specify the schema with AS.

2

Filter sales for the 'Electronics' category

Create a new relation called electronics_sales by filtering sales to keep only records where category equals 'Electronics'. Use the FILTER statement.

Hadoop

sales = LOAD 'sales_data.csv' USING PigStorage(',') AS (product:chararray, category:chararray, amount:int);
-- Filter sales where category is 'Electronics'
-- Your code here

Hint

Use FILTER with the condition category == 'Electronics'.

3

Group sales by product and calculate total amount

Group electronics_sales by product into a relation called grouped_sales. Then, create a relation called total_sales that calculates the sum of amount for each product using the FOREACH and GENERATE statements.

Hadoop

sales = LOAD 'sales_data.csv' USING PigStorage(',') AS (product:chararray, category:chararray, amount:int);
electronics_sales = FILTER sales BY category == 'Electronics';
-- Group electronics_sales by product
-- Calculate total amount per product
-- Your code here

Hint

Use GROUP to group by product, then FOREACH ... GENERATE with SUM to calculate totals.

4

Display the total sales per product

Use the DUMP statement to display the contents of total_sales.

Hadoop

sales = LOAD 'sales_data.csv' USING PigStorage(',') AS (product:chararray, category:chararray, amount:int);
electronics_sales = FILTER sales BY category == 'Electronics';
grouped_sales = GROUP electronics_sales BY product;
total_sales = FOREACH grouped_sales GENERATE group AS product, SUM(electronics_sales.amount) AS total_amount;
-- Display the total_sales relation
-- Your code here

Hint

Use DUMP to print the relation contents.

Practice

1. Why is Pig considered simpler for data transformation compared to writing raw MapReduce code in Hadoop?

easy

A. Because Pig uses a high-level scripting language that is easier to write and understand

B. Because Pig requires writing Java code for every task

C. Because Pig only works with small datasets

D. Because Pig does not support data filtering or grouping

Solution

1. Step 1: Understand Pig's language level

   Pig uses Pig Latin, a high-level scripting language designed to simplify data transformations.

2. Step 2: Compare with MapReduce complexity

   MapReduce requires writing detailed Java code, which is more complex and verbose than Pig Latin scripts.

3. Final Answer:

   Because Pig uses a high-level scripting language that is easier to write and understand -> Option A

4. Quick Check:

   High-level scripting = Simplicity [OK]

Hint: Pig Latin scripts are simpler than Java MapReduce code [OK]

Common Mistakes:

• Thinking Pig requires Java coding
• Assuming Pig only works on small data
• Believing Pig lacks filtering features

2. Which of the following is the correct syntax to load comma-delimited data from a file named data.txt in Pig?

easy

A. LOAD 'data.txt' USING PigStorage();

B. load data.txt with PigStorage(',');

C. LOAD 'data.txt' USING PigStorage(',');

D. LOAD data.txt USING PigStorage();

Solution

1. Step 1: Recall Pig LOAD syntax

   The correct syntax requires the filename in quotes and specifying the loader function with parameters.

2. Step 2: Check each option

   LOAD 'data.txt' USING PigStorage(','); correctly uses quotes around filename and specifies PigStorage with delimiter ','. Options B and C have syntax errors or missing quotes. LOAD 'data.txt' USING PigStorage(); misses the delimiter argument.

3. Final Answer:

   LOAD 'data.txt' USING PigStorage(','); -> Option C

4. Quick Check:

   Quotes + PigStorage with delimiter = Correct LOAD [OK]

Hint: Filename must be in quotes and PigStorage needs delimiter [OK]

Common Mistakes:

• Omitting quotes around filename
• Not specifying delimiter in PigStorage
• Using lowercase LOAD keyword

3. Given the Pig script:

A = LOAD 'data.txt' USING PigStorage(',') AS (name:chararray, age:int);
B = FILTER A BY age > 25;
C = GROUP B BY name;
D = FOREACH C GENERATE group, COUNT(B);
DUMP D;

What will be the output if data.txt contains:

John,30
Jane,22
John,28
Jane,26

medium

A. (John,2)
(Jane,2)

B. (John,3)
(Jane,2)

C. (John,1)
(Jane,1)

D. (John,2)
(Jane,1)

Solution

1. Step 1: Filter rows where age > 25

   From data, John(30), John(28), and Jane(26) pass the filter; Jane(22) is excluded.

2. Step 2: Group by name and count

   John has 2 records, Jane has 1 record after filtering.

3. Final Answer:

   (John,2) (Jane,1) -> Option D

4. Quick Check:

   Filter >25 then count per name = (John,2), (Jane,1) [OK]

Hint: Filter first, then group and count [OK]

Common Mistakes:

• Counting all records without filtering
• Mixing up group and count results
• Ignoring filter condition

4. Identify the error in this Pig script:

A = LOAD 'data.txt' USING PigStorage(',') AS (name:chararray, age:int);
B = FILTER A BY age > '25';
DUMP B;

medium

A. PigStorage cannot be used with comma delimiter

B. Using quotes around number in filter condition causes a type error

C. Missing semicolon after LOAD statement

D. AS clause cannot define schema

Solution

1. Step 1: Check filter condition data types

   The filter compares age (int) with '25' (string), causing a type mismatch error.

2. Step 2: Validate other syntax elements

   Semicolons are present, PigStorage supports comma delimiter, and AS clause is valid for schema definition.

3. Final Answer:

   Using quotes around number in filter condition causes a type error -> Option B

4. Quick Check:

   Number in filter must be unquoted integer [OK]

Hint: Filter numbers without quotes to avoid type errors [OK]

Common Mistakes:

• Quoting numeric values in filters
• Assuming missing semicolon error
• Misunderstanding PigStorage usage

5. You have a dataset with user purchases: (user, item, price). You want to find the total spending per user but only for items priced above 100. Which Pig script sequence correctly achieves this?

hard

A. A = LOAD 'purchases.txt' USING PigStorage(',') AS (user:chararray, item:chararray, price:int);
B = FILTER A BY price > 100;
C = GROUP B BY user;
D = FOREACH C GENERATE group, SUM(B.price);
DUMP D;

B. A = LOAD 'purchases.txt' USING PigStorage(',') AS (user:chararray, item:chararray, price:int);
B = GROUP A BY user;
C = FILTER B BY price > 100;
D = FOREACH C GENERATE group, SUM(B.price);
DUMP D;

C. A = LOAD 'purchases.txt' USING PigStorage(',') AS (user:chararray, item:chararray, price:int);
B = FILTER A BY price > 100;
C = FOREACH B GENERATE user, SUM(price);
DUMP C;

D. A = LOAD 'purchases.txt' USING PigStorage(',') AS (user:chararray, item:chararray, price:int);
B = FOREACH A GENERATE user, item, price;
C = FILTER B BY price > 100;
DUMP C;

Solution

1. Step 1: Filter items with price > 100

   A = LOAD 'purchases.txt' USING PigStorage(',') AS (user:chararray, item:chararray, price:int); B = FILTER A BY price > 100; C = GROUP B BY user; D = FOREACH C GENERATE group, SUM(B.price); DUMP D; filters dataset first to keep only items priced above 100.

2. Step 2: Group filtered data by user and sum prices

   Then it groups by user and calculates total spending using SUM on price.

3. Step 3: Check other options

   A = LOAD 'purchases.txt' USING PigStorage(',') AS (user:chararray, item:chararray, price:int); B = GROUP A BY user; C = FILTER B BY price > 100; D = FOREACH C GENERATE group, SUM(B.price); DUMP D; filters after grouping (invalid), C tries SUM without grouping, D only filters without aggregation.

4. Final Answer:

   A = LOAD 'purchases.txt' USING PigStorage(',') AS (user:chararray, item:chararray, price:int); B = FILTER A BY price > 100; C = GROUP B BY user; D = FOREACH C GENERATE group, SUM(B.price); DUMP D; -> Option A

5. Quick Check:

   Filter then group then sum = Correct approach [OK]

Hint: Filter before grouping to sum correct subset [OK]

Common Mistakes:

• Filtering after grouping
• Summing without grouping
• Not filtering before aggregation