Skip to content

Financial Management and Business Data Analytics · Data Processing, Organisation, Cleaning and Validation

Data Transformation, Integration and Preparation for CMA Inter

Updated 10 October 2026 · Fact-checked

Data transformation converts data into a consistent format. Data integration merges data from several sources into one view. ETL means Extract, Transform, Load: pull data from sources, clean and reshape it, then store it in a target system. Data preparation is the full set of steps that makes data ready for analysis.

Understand Data Transformation, Integration and Preparation

Business data rarely arrives ready to use. Sales sit in one system, payroll in another, and bank data comes as spreadsheets. Each source may use different formats, units, date styles and codes. You cannot analyse such data until it is made consistent and combined.

Data transformation changes data from its raw form into a form suited for analysis. Common transformations are converting data types (text to number, text to date), standardising formats (DD-MM-YYYY everywhere), unifying units (₹ lakh vs ₹ crore), deriving new fields (profit = revenue − cost), aggregating (daily to monthly), and encoding categories (Yes/No to 1/0).

Data integration combines data from multiple sources into one unified dataset. You match records using a common key, such as customer ID or invoice number. Typical issues are duplicate records, different names for the same field, and conflicting values for the same item. These must be resolved before or during merging.

ETL is the standard process for this. Extract: read data from source systems (databases, files, applications). Transform: clean, validate, convert and reshape it. Load: write it into a target such as a data warehouse. Data is usually staged first, so errors do not damage the target.

Two scaling ideas are often confused. Normalisation (min-max scaling) rescales values into a fixed range, usually 0 to 1. Standardisation (z-score) rescales values so the mean is 0 and the standard deviation is 1. Both make variables with different units comparable. Note that in database design, normalisation can also mean organising tables to reduce redundancy. Read the question to see which meaning is used.

Key rules to remember

Min-max normalisation
x′ = (x − min) ÷ (max − min)
Gives values from 0 to 1. Sensitive to outliers because min and max change.
Standardisation (z-score)
z = (x − μ) ÷ σ
μ is the mean and σ the standard deviation. Result has mean 0 and standard deviation 1. Not limited to a fixed range.
ETL sequence
Extract → Transform → Load
Order matters: data is transformed before it is loaded into the target. In ELT, loading happens before transformation.
Data preparation flow
Collect → Integrate → Clean → Transform → Validate → Load/Use
A practical order for writing answers. Textbooks may group or name steps slightly differently.

How to solve Data Transformation, Integration and Preparation questions

Use this method for definition, process and small numerical questions on this topic.

  1. 1Identify what is asked: a definition, the ETL stages, a preparation sequence, or a scaling calculation.
  2. 2State the key term in one line, for example what transformation or integration means.
  3. 3List the stages or techniques in the correct order and explain each in one line with a business example.
  4. 4For a scaling question, write the formula first, then find min, max, mean and standard deviation from the data given.
  5. 5Substitute carefully and compute each value. Check that normalised values lie between 0 and 1.
  6. 6Add a one-line conclusion on why this makes data reliable and comparable for analysis.

Quickest way: ETL and scaling shortcut

When to use it: Use for MCQs and short notes when time is tight.

  1. For ETL, ask: is data being read (Extract), changed (Transform) or stored (Load)?
  2. For scaling, ask: does the result lie within 0 to 1? If yes, it is normalisation. If it is centred on 0, it is standardisation.
  3. For normalisation, the minimum value always becomes 0 and the maximum becomes 1. Use this to check answers.
  4. For standardisation, a value equal to the mean always gives z = 0.

Common mistakes in Data Transformation, Integration and Preparation

  • Treating normalisation and standardisation as the same thing.

    Both rescale data and the names sound alike.

    Fix: Normalisation maps to a fixed range such as 0 to 1 using min and max. Standardisation uses mean and standard deviation to give mean 0 and SD 1.

  • Writing the ETL stages in the wrong order or mixing up Transform and Load.

    Students memorise the letters without the meaning.

    Fix: Remember the flow: take data out, fix it, put it in. Cleaning and conversion belong to Transform.

  • Saying data integration is just copying files together.

    The word merge sounds simple.

    Fix: Mention matching on a common key, removing duplicates, resolving conflicts and aligning formats.

  • Using max − min wrongly in min-max scaling, such as dividing by max alone.

    Rushing under time pressure.

    Fix: Always write (x − min) ÷ (max − min). Check that the smallest value gives 0.

  • Skipping validation after transformation.

    Students think the job ends once data is converted.

    Fix: Add checks such as record counts, totals matching the source, and range checks before loading.

Worked examples

Example 1

Monthly sales (₹ lakh) of a firm for five branches are 20, 30, 40, 50 and 60. Normalise the value 40 using min-max scaling, and also normalise 60.

Show the solution
  1. Minimum = 20 and maximum = 60, so max − min = 40.
  2. For 40: (40 − 20) ÷ 40 = 20 ÷ 40 = 0.5.
  3. For 60: (60 − 20) ÷ 40 = 40 ÷ 40 = 1.
  4. Check: the maximum gives 1, so the method is applied correctly.

Answer: Normalised value of 40 is 0.5 and of 60 is 1.

Example 2

Explain the ETL process for a company that combines sales data from its billing software and expense data from spreadsheets into one reporting database. Then standardise a branch sales figure of ₹70 lakh when the mean is ₹50 lakh and the standard deviation is ₹10 lakh.

Show the solution
  1. Extract: read invoice data from the billing software and expense data from the spreadsheets without changing the sources.
  2. Transform: convert dates to one format, express amounts in the same unit, remove duplicate invoices, fix missing values, and match records on a common key such as branch code.
  3. Validate: check that total sales after transformation equal the source totals.
  4. Load: write the cleaned, integrated data into the reporting database for analysis.
  5. Standardisation: z = (x − μ) ÷ σ = (70 − 50) ÷ 10 = 2.
  6. Interpretation: the branch sales are 2 standard deviations above the mean.

Answer: ETL extracts from both sources, transforms and validates the data, then loads it into the reporting database. The z-score of ₹70 lakh is 2.

Exam tips

  • Learn the ETL definitions as three one-line points and give a business example for each.
  • For numericals, always write the formula before substituting, because step marks are awarded.
  • In MCQs, look for the range clue: 0 to 1 means normalisation; mean 0 and SD 1 means standardisation.
  • In theory answers, link integration to key matching, duplicate removal and consistent formats.
  • Keep this topic separate from cleaning and validation: here the focus is conversion, merging and readiness for analysis.

Practice questions from Data Processing, Organisation, Cleaning and Validation

Data Transformation, Integration and Preparation: frequently asked questions

What is ETL in data analytics?

ETL stands for Extract, Transform, Load. Data is extracted from source systems, cleaned and reshaped in the transform stage, and then loaded into a target such as a data warehouse. It gives analysts consistent, reliable data in one place.

What is the difference between normalisation and standardisation?

Normalisation rescales values to a fixed range, usually 0 to 1, using the minimum and maximum. Standardisation converts values to z-scores with mean 0 and standard deviation 1. Normalisation is affected more by outliers because it depends on the extremes.

How is data transformation different from data integration?

Transformation changes the format, type or scale of data. Integration combines data from different sources into one dataset. In practice they happen together, as data is often transformed so that it can be integrated.

What are the main steps of data preparation?

Collect the data, integrate sources, clean errors and duplicates, transform formats and scales, and validate the result. Then the data is ready for analysis. State the steps in this logical order in your answer.