Data Types, Processing & Business Applications

TAKEAWAYS:

• Data is the raw material for information, and its type (numeric, alphanumeric, multimedia) determines how it’s stored and processed.
• The data processing cycle (input → processing → storage → output) is the backbone of all IT systems, from eSewa transactions to Daraz order fulfillment.
• Data mining and data warehousing turn raw data into actionable insights (e.g., Ncell predicting customer churn).
• File processing vs. database processing shows why modern systems (like NEPSE’s stock data) use DBMS for efficiency and scalability.
• Data validation and integrity ensure accuracy in critical systems (e.g., Khalti’s transaction records).
• Ethical and legal considerations (privacy, security) apply to all data handling, from Pathao’s rider data to NTC’s network logs.

What is Data?

Data is the raw, unprocessed facts and figures collected for analysis. It lacks context or meaning until processed. For example:

• Raw data: "5000", "Kathmandu", "2023-10-15" (from a Daraz order).
• Processed information: "Order #5000 from Kathmandu delivered on 2023-10-15 with a delay of 2 hours."

Types of Data

Data is classified based on format, structure, and usage. Here’s a comparison table:

How Data is Processed: The Data Processing Cycle

The data processing cycle converts raw data into useful information through input → processing → storage → output. Visualize it as a loop:

graph TD
    A["Input: Raw Data"] --> B["Processing: Operations"]
    B --> C["Storage: Databases/Files"]
    C --> D["Output: Information"]
    D --> A

Steps Explained with a Real Example: eSewa Transaction

1. Input:

   • User enters numeric data (₹500), alphanumeric data (phone number "98XXXXXXXX"), and textual data (transaction note "Rent").
   • Multimedia: QR code scan (binary data).
   • Geospatial: Location of the merchant (for fraud detection).

2. Processing:

   • Validation: Checks if the phone number is registered (data integrity).
   • Calculation: Adds transaction fee (₹500 + 2% = ₹510).
   • Encryption: Converts data to binary for secure transfer (binary data).
   • Database Update: Records the transaction in eSewa’s DBMS (storage).

3. Storage:

   • Stored in a relational database (tables for users, transactions, merchants).
   • Backup: Cloud storage for disaster recovery.

4. Output:

   • Confirmation SMS: "Transaction successful. Debited ₹510 from 98XXXXXXXX."
   • Merchant Alert: "New payment received: ₹500 (fee: ₹10)."
   • Analytics Dashboard: eSewa’s backend shows transaction trends (data mining).

Data Processing Methods

1. Manual Processing

• Definition: Data handled by humans (no IT tools).
• Example: A shopkeeper writing sales in a notebook.
• Disadvantages:
  • Slow, error-prone.
  • No scalability (e.g., NTC manually tracking network issues in 2000s).
• When Used: Small businesses, temporary records.

2. Mechanical Processing

• Definition: Uses machines (typewriters, calculators) but no computers.
• Example: Old cash registers in local kirana shops.
• Limitations:
  • No data storage or analysis.
  • Prone to physical damage (e.g., a broken calculator in a bank).

3. Electronic Processing

• Definition: Uses computers and software for speed and accuracy.
• Example: Banks processing loans using software (e.g., NMB’s mortgage calculator).
• Advantages:
  • Speed: Processes 1000s of transactions/sec (e.g., Khalti’s UPI system).
  • Accuracy: Reduces human errors (e.g., NEPSE’s automated trading).
  • Storage: Cloud databases (e.g., Daraz’s inventory system).

Data Processing Techniques

Data Models

A data model defines how data is structured, stored, and accessed. Three key types:

1. Hierarchical Model

   • Structure: Tree-like (parent-child relationships).
   • Example: Old file systems where a "root" folder contains subfolders.
   • Limitation: Inflexible (e.g., adding a new category requires restructuring).

   graph TD
       A["Root: Customers"] --> B["Child: Retail"]
       A --> C["Child: Wholesale"]
       B --> D["Grandchild: Kathmandu"]
       B --> E["Grandchild: Pokhara"]

2. Network Model

   • Structure: Multiple parent-child relationships (more flexible than hierarchical).
   • Example: NTC’s network topology where a tower connects to multiple users.

3. Relational Model

   • Structure: Tables (rows = records, columns = fields) linked via keys.
   • Example: Daraz’s database with tables for Users, Products, and Orders.
   • Advantage: Supports complex queries (e.g., "Find all orders from Pokhara in October").

File Processing vs. Database Processing

In the Real World

1. eSewa’s Transaction Processing

   • Idea Used: Data validation and relational database processing.
   • How: When you pay via eSewa:
     • The app validates your phone number (data integrity).
     • The transaction is recorded in a relational database linking your account, the merchant, and the payment.
     • Data mining helps eSewa detect fraud (e.g., unusual transaction locations).

2. Pathao’s Rider Management

   • Idea Used: Geospatial data + real-time processing.
   • How:
     • Riders’ locations are tracked via GPS coordinates (geospatial data).
     • The app processes this data to match riders with nearby passengers (optimization algorithm).
     • Multimedia: Profile pictures and ID scans (binary data) verify riders.

3. NTC’s Network Optimization

   • Idea Used: Data aggregation and filtering.
   • How:
     • NTC collects network logs (data input) from towers across Nepal.
     • Filtering: Identifies outages in specific areas (e.g., Kathmandu’s Thapathali).
     • Aggregation: Calculates average latency to improve service (e.g., "4G speed dropped by 20% in Bhaktapur").
     • Output: Engineers use this data to upgrade infrastructure.

Data Mining: Turning Data into Gold

Definition: The process of discovering hidden patterns, correlations, or trends in large datasets using statistical methods, AI, and database systems.

Steps in Data Mining

flowchart TD
    A["Data Collection"] --> B["Data Cleaning"]
    B --> C["Data Integration"]
    C --> D["Data Transformation"]
    D --> E["Data Mining: Pattern Discovery"]
    E --> F["Pattern Evaluation"]
    F --> G["Knowledge Presentation"]

Real-World Example: Ncell’s Churn Prediction

• Problem: Ncell loses 10% of customers yearly due to competitors like NTC.
• Solution: Data mining to predict which customers might leave.
  • Data Collected:
    • Call duration, SMS usage, payment history (numeric).
    • Customer complaints (textual).
    • Location data (geospatial).
  • Analysis:
    • Customers who reduce call minutes by 30% in a month are 60% likely to churn.
    • Those who switch SIMs frequently are high-risk.
  • Output:
    • Ncell sends discounted plans to at-risk customers.
    • Result: Churn rate drops by 15%.

Applications in Nepal

Data Warehousing vs. Data Mining

Challenges in Data Processing

1. Data Quality Issues

   • Problem: Incomplete or incorrect data (e.g., a Daraz order with no delivery address).
   • Solution: Data validation rules (e.g., "Address field cannot be empty").

2. Security Risks

   • Problem: Data breaches (e.g., Khalti hack in 2021 exposed user data).
   • Solution:
     • Encryption (binary data protection).
     • Access controls (only authorized staff can view sensitive data).

3. Privacy Concerns

   • Problem: GDPR-like laws are emerging in Nepal (e.g., Digital Transaction Act, 2018).
   • Example: NTC cannot share customer call logs without consent.

4. Scalability

   • Problem: Systems slow down with growth (e.g., Pathao’s app crashing during Diwali).
   • Solution: Cloud databases (e.g., Daraz uses AWS to handle Black Friday traffic).

Exam Tip

This unit is conceptual but heavily tested in TU/PU exams. Focus on:

1. Definitions:

   • Know the exact definitions of data, information, data mining, and data model.
   • Example: "Data mining is the process of discovering meaningful patterns from large datasets using techniques like classification, clustering, and association."

2. Comparisons:

   • Be ready to compare file processing vs. database processing or hierarchical vs. relational models in tables.
   • Example question: "Why do modern businesses prefer DBMS over file processing?" → Answer with redundancy, integrity, and scalability.

3. Real-World Applications:

   • Always tie answers to Nepali examples. For instance:
     • "How does eSewa use data processing?" → Input (user data) → Processing (validation) → Output (SMS confirmation).
   • Diagrams save marks: Draw the data processing cycle or a relational database schema in exams.

4. Short-Answer Questions:

   • Practice 3-mark questions like:
     • "Mention three advantages of database processing over file processing." Answer:
       1. Reduced redundancy (data stored once).
       2. Improved data integrity (constraints prevent errors).
       3. Better security (access controls).
     • "Define data mining with an example." Answer: "Data mining is extracting hidden patterns from large datasets. Example: Ncell uses it to predict customer churn by analyzing call duration trends."

5. Case Studies:

   • Exams may ask: "How would you design a data processing system for a bank like NMB?" Structure your answer:
     1. Input: Customer details (numeric/alphanumeric).
     2. Processing: Loan eligibility checks (validation rules).
     3. Storage: Relational database with tables for Customers, Loans, Payments.
     4. Output: Loan approval/rejection letters + analytics dashboard.

Worked Example: Calculating Loan Interest (NMB Bank)

Scenario: Mr. Thapa takes a ₹500,000 loan at 8% annual interest, repayable in 5 years. Calculate the monthly installment using the fixed installment method.

Step 1: Understand the Data

• Principal (P): ₹500,000 (numeric data).
• Annual Interest Rate (r): 8% → Monthly rate = $r=\frac{8}{100\times 12}=0.006667$.
• Loan Term (n): 5 years × 12 months = 60 months.

Step 2: Formula for Fixed Installment (EMI)

The monthly payment $M$ is calculated using: $$M=P\times \frac{r(1+r{)}^n}{(1+r{)}^n-1}$$

Step 3: Plug in the Values

$$M=500,000\times \frac{0.006667\times (1.006667{)}^{60}}{(1.006667{)}^{60}-1}$$ $$M=500,000\times \frac{0.006667\times 1.4889}{1.4889-1}$$ $$M=500,000\times \frac{0.009926}{0.4889}\approx 500,000\times 0.020298\approx ₹10,149$$

Step 4: Data Processing in the Bank’s System

1. Input: Loan application data (numeric: amount, rate; textual: customer name).
2. Processing:
   • The bank’s software applies the EMI formula (automated calculation).
   • Validation: Checks if Mr. Thapa’s income (₹80,000/month) meets the 40% EMI-to-income rule (₹10,149 ≤ 32,000).
3. Storage: Loan details stored in the bank’s DBMS (tables: Loans, Payments).
4. Output:
   • Loan agreement: "Monthly installment: ₹10,149 for 60 months."
   • Analytics: Bank uses this data to assess risk (e.g., "Loans > ₹500K have a 5% default rate").

Key Takeaways for Exams

1. Memorize the data processing cycle and label it in diagrams.
2. Know the differences between file processing and DBMS (use tables).
3. Relate concepts to Nepali apps:
   • eSewa → data validation + relational DB.
   • Daraz → data mining for recommendations.
   • NTC → geospatial data for network optimization.
4. Practice numerical questions (e.g., EMI calculations, sorting algorithms).
5. Ethics matter: Always mention privacy and security in answers involving real-world data (e.g., Khalti’s encryption).