3  Measurement

Jab hum data science ki baat karte hain, to pemaish ya measurement ka role bohot ahem hota hai. Is chapter mein, hum dekheinge ke kaise data science mein pemaish ke mukhtalif pehlu hote hain aur ye kyun zaroori hai. 📏📊

3.1 Importance of Measurement

Data science mein, measurement ka role bohot ahem hota hai. Agar hum kisi cheez ko measure nahi kar sakte, to hum us cheez ko analyze bhi nahi kar sakte. Is liye, measurement bohot zaroori hai.

Data ki Sifaat ka Taayun (Determining Data Quality):

Measurement se hum data ki accuracy, reliability, aur validity ka taayun karte hain. Yeh samajhna zaroori hai ke aap jo data use kar rahe hain wo qabile bharosa aur durust hai. 📊🔬

Misal ke taur par, agar aap ek research kar rahe hain jisme aap logon se unke khane ke adat ke bare mein sawalat pooch rahe hain. Yahan, aapko yeh dekhna hoga ke jawabat kitne sahi hain, kya log sach bol rahe hain, ya unke jawabat mein kuch bias to nahi hai. 🍽️📋

Measurement Scales aur Unka Istemal:

Har type ka data alag tarah se measure kiya jata hai aur iske liye alag scales hoti hain. Ye scales hain: nominal, ordinal, interval, aur ratio. Har ek ka apna unique faida aur istemal hai. 📐📏

Jaise, nominal scale mein hum cheezon ko naam se pehchante hain, masalan, kisi survey mein mard ya aurat ke options. Ordinal scale mein, hum order ya darja bandi karte hain, jaise, hotel reviews mein stars. Interval scale mein koi fix zero point nahi hota, masalan, temperature. Aur ratio scale mein fixed zero point hota hai, jaise, kisi cheez ka wazan ya lambai. ⚖️🌡️

3.2 Scales/Levels of Measurement 📐📊

Pemayesh Ke Paimane ya Darje

Measurement scales, ya pemayesh ke paimane, data science mein data ko categorize aur analyze karne ka ek basic framework faraham karte hain. Har scale data ki mukhtalif kism ke properties ko measure karta hai aur iska apna unique use hota hai.

3.2.1 Nominal Scale (Nam Ka Paimana):

• Definition: Nominal scale sab se basic level ka measurement scale hai. Is mein data ko categories mein divide kiya jata hai, lekin in categories mein koi numeric order ya value nahi hoti.
• Example: Jaise, a survey mein logon ki nationality ya unka profession poocha jata hai. Pakistani, Indian, Teacher, Doctor, etc., are examples of nominal data.
• Use in Data Science: Data sorting aur categorization ke liye istemal hota hai, jaise customer segmentation ya demographic studies mein. 🌍👥

3.2.2 Ordinal Scale (Tarteebi Paimana):

• Definition: Ordinal scale mein, data categories mein hota hai, lekin in categories mein ek specific order ya sequence hoti hai.
• Example: Jaise, ek survey mein logon se unki education level ke bare mein poocha jata hai: Matric, Intermediate, Bachelor’s, Master’s. Yahan, har category ka ek specific order hai.
• Use in Data Science: Data ko rank ya order mein rakhne ke liye istemal hota hai, masalan, customer satisfaction surveys mein. 😃📊

3.2.3 Interval Scale (Waqt Ke Faslay Ka Paimana):

• Definition: Interval scale numeric values ke sath aata hai, aur is mein equal intervals ya differences hote hain, lekin iska koi true zero point nahi hota.
• Example: Temperature Celsius ya Fahrenheit mein. Yahan, 0 degrees ka matlab ye nahi ke koi temperature nahi hai; ye sirf ek point hai scale par.
• Use in Data Science: Data mein variations ko samajhne aur analyze karne ke liye, jaise climate change studies. 🌡️🌏

3.2.4 Ratio Scale (Tanasubi Paimana):

• Definition: Ratio scale interval scale ki tarah hota hai lekin is mein ek absolute zero point hota hai.
• Example: Distance (meters ya kilometers mein), weight (kilograms), ya age (saal mein). Yahan, zero ka matlab hai ke us cheez ka non-existence hai.
• Use in Data Science: Quantitative analysis aur scientific calculations ke liye, jaise physics ya engineering applications. 🔬⚖️

Outline

Is section mein, measurement ke mukhtalif scales ya darje aur unke data science mein istemal ko tafseel se samjhaya gaya hai. Har scale ke unique features aur examples ko include kiya gaya hai, taake readers ko clear understanding ho ke kaise ye scales data ko samajhne aur analyze karne mein madadgar hain. Ye section data science practitioners ke liye important hai kyun ke ye unhe guide karta hai ke kis tarah ke data ko kaise handle kiya jaye aur kis tarah ke analysis ke liye konsa scale behtar hai.

A tabulated form to describe the scales of measurement in detail:

Scale Type	Definition	Examples	Usage in Data Science
Nominal Scale	Categories without any numeric order. Differentiates by type, not quantity or order.	Gender, Nationality, Occupation	Used for categorizing and segmenting data, like in customer segmentation or demographic studies.
Ordinal Scale	Categories with a specific order or sequence, but the intervals are not necessarily equal.	Education Level, Satisfaction Ratings	Used for ranking or ordering data, like in customer satisfaction surveys or educational qualifications.
Interval Scale	Numeric scale with equal intervals between values, but no true zero point.	Temperature (Celsius/Fahrenheit), Calendar Years	Used for measuring differences and averages in data, like in climate studies or historical timelines.
Ratio Scale	Similar to interval scale but with a true zero point, allowing for statements of magnitude.	Weight, Height, Age, Distance	Used for comprehensive quantitative analysis and scientific calculations, like in physics or engineering applications.

3.3 Data Collection and Measurement

Data Collection ke Process (The Process of Data Collection):

Data Collection Techniques: Data science mein data ikhatta karne ke mukhtalif tareeqe hote hain, jaise surveys, experiments, aur field studies. Har technique ka apna unique maqsad aur faida hota hai. 📋🔍

Misal ke taur par, agar aap market research kar rahe hain to aap online surveys ya focus groups ka istemal kar sakte hain. Ye aapko tezi se aur wasee range mein data faraham karta hai. Ya phir, agar aap environmental studies kar rahe hain, to field observations aur experiments zyada munasib ho sakte hain. 🌳🧑‍🔬

Measurement Errors ki Samajh (Understanding Measurement Errors):

Common Errors: Data collection process mein aane wale common errors mein shamil hain sampling error, bias, aur data entry mistakes. Ye errors aapke data ke results ko significantly affect kar sakte hain. ⚠️🚫

Jaise, agar aap ek survey mein sirf aik khas age group ke logon ko include karte hain, to ye sampling bias create kar sakta hai. Ya phir, data entry mein ghalti se galat information enter ho jaye, to ye bhi results ko distort kar sakta hai. 💻📉

Errors ko Kam Karna (Minimizing Errors):

Strategies to Reduce Errors: Kuch strategies jin se aap errors ko kam kar sakte hain, jaise careful planning, diverse sampling, aur data verification processes. Is se aapka data zyada reliable aur accurate banega. 📈✅

Misal ke taur par, aap pehle se hi decide kar lein ke aapki sample population kaisi hogi, taake aapke data mein diversity ho. Aur data collection ke baad, aap data verification aur cleaning process se guzar kar kisi bhi possible errors ko identify aur correct kar sakte hain. 🧹🔧

3.4 Operationalization and Proxy Measures

3.4.1 Operationalization (Amliyat ka Tareeqa-kar) 📋🔧

Operationalization, ya amliyat ka tareeqa-kar, ek research process ka hissa hai jisme complex concepts ko measurable form mein tabdeel kiya jata hai. 📐🧠

• Tafseel: Jab hum research karte hain, to kai dafa humein abstract concepts (jaise khushi, ghurbat, ya sehat) ko quantify karna parta hai. Operationalization is process ko kehte hain jisme hum in concepts ko aise variables mein convert karte hain jo hum measure kar sakein. 🌐💭

Misal ke taur par, agar aap “khushi” ko measure karna chahte hain, to aap isay various indicators jaise life satisfaction, positive experiences, ya smile frequency ke through measure kar sakte hain. 😀📊

• Application: Operationalization research design mein crucial hai kyun ke yeh humein specific, measurable, aur quantifiable data provide karta hai jo humare conclusions aur analysis ko more credible banata hai. 📝✅

3.4.2 Proxy Measurement (Proxy Pemayesh) 📏🔍

Proxy measurement, ya proxy pemayesh, tab istemal hoti hai jab direct measurement mushkil ya na-mumkin ho. 🚧📈

• Tafseel: Proxy measurement ek ‘stand-in’ ya alternate measurement hoti hai jo asal variable ki jagah use ki jati hai. Yeh tab kiya jata hai jab asal variable ko direct measure karna mushkil ho. 🔄🔗

Jaise, agar aap kisi mulk ki economic health measure karna chahte hain, to direct isay measure karna mushkil hai. Is ki jagah, aap GDP growth rate, unemployment rate, ya consumer spending jaise indicators ka istemal kar sakte hain as proxies. 💹💰

• Application: Proxy measurements research mein common hain, khaas tor par social sciences aur economics mein, jahan direct measurement ke liye resources ya access limited ho. Ye technique humein phir bhi important insights provide karti hai, albeit with some level of assumption or indirectness. 🌍🔑

3.5 Surrogate Endpoints 🎯🔬

Surrogate Endpoints, ya mutaabadil anjaam ke nuqaat, medical research aur clinical trials mein istemal hone wale aise markers hain jo barah-e-raast bemari ke anjaam ko naapne ke bajaye uske effects ya risk factors ko measure karte hain. 🩺📊

• Tafseel: Yeh aksar un halaton mein istemal hota hai jahan asal clinical endpoint (jaise, marz ki rok-thaam ya ilaj ki kamyabi) ko measure karna mushkil ho ya bohot waqt le. Surrogate endpoints se researchers ko jaldi aur aasaani se samajhne mein madad milti hai ke aik treatment ya dawa kitni effective hai. 🚑💊

Misal ke taur par, agar ek nai dawai ka test kiya ja raha hai jo cholesterol ko kam karta hai, to researchers direct heart attacks ya strokes ki kami ko naapne ke bajaye cholesterol levels ko measure karte hain as a surrogate endpoint. Ye assumption yeh hota hai ke kam cholesterol level se heart attacks ka risk bhi kam ho jata hai. ❤️📉

• Application: Surrogate endpoints zyada tar chronic diseases (jaise diabetes, hypertension) ke research mein istemal hote hain. Ye researchers ko enable karta hai ke wo tezi se aur kam resources ke sath potential treatments ki efficacy ko samjhein aur evaluate karein. 📝🔎

• Ehmiyat aur Tanqeed: Surrogate endpoints ka istemal time aur resources ki bachat to karta hai, lekin iska istemal kabhi-kabhi misleading bhi ho sakta hai. Agar surrogate endpoint aur asal health outcome ke darmiyan strong relationship na ho, to is se galat conclusions nikal sakte hain. Is liye, in endpoints ka chayan aur interpretation bohot soch-samajh ke aur scientific evidence ke sath karna chahiye. 🤔💡

3.6 Quantitative and Qualitative Measurement 📊📖

Quantitative Data ki Tafseel (Detailing Quantitative Data):

• Definition and Examples: Quantitative data wo hota hai jo numbers mein measure kiya ja sakta hai. Is mein typically counts, percentages, ya numerical values shamil hain. 📉🔢

  Misal ke taur par, ek company ki monthly sales, ek website par rozana ke visitors, ya kisi school ke students ke exam scores. Ye data humein concrete aur measurable information deta hai, jaise kitna, kitni baar, aur kis darje mein.

Qualitative Data ka Analysis (Analyzing Qualitative Data):

• Nature and Interpretation: Qualitative data non-numeric hota hai aur ismein text, images, ya observations shamil hote hain. Is data ko samajhna aur interpret karna often zyada complex hota hai. 📚🎨

  Jaise, customer reviews, interview transcripts, ya observational notes. Ye data humein deeper insights deta hai jaise log kya sochte hain, kyun kisi cheez ko pasand ya napasand karte hain, aur unke experiences kaise hote hain.

Combining Quantitative and Qualitative Data (Dono Types ke Data ko Milana):

• Hybrid Approach: Behtareen insights often dono types ke data ko combine kar ke milte hain. Is approach se hum both measurable outcomes aur deeper human experiences ko samajh sakte hain. 🤝📊

Ek retail store ka misal lein: Store quantitative data se sales trends aur popular items ko track karta hai, jabke customer interviews aur feedback se ye samajhne ki koshish karta hai ke customers kyun kisi product ko prefer karte hain ya unke shopping experience mein kya behtar kiya ja sakta hai.

3.7 Data and Types of Data 📊📈

What is Data?

Data is the raw material of data science. It is the information that we collect and analyze to gain insights and make decisions. Data can be quantitative or qualitative, and it can be collected through various methods, like surveys, experiments, or field studies. Data is the foundation of data science, and it is the basis of all data science processes and techniques.

Is liye, data science mein data ki ahmiyat bohot zyada hai. Is chapter mein, hum dekheinge ke data kya hota hai aur data ki mukhtalif types kya hain. 📊📈

Primary and Secondary data are two fundamental categories based on the source and nature of the data collection process.

3.7.1 Primary Data vs. Secondary Data

3.7.1.1 Primary Data

• Definition: Primary data is data collected directly by the researcher for the specific purpose of their study. It is original and collected at the source.
• Methods of Collection: Includes surveys, interviews, experiments, questionnaires, observations, and focus groups.
• Examples:
  • A researcher conducting a survey to study consumer behavior.
  • Field experiments in environmental studies.
• Uses:
  • Tailored to the specific needs and questions of the research.
  • Provides up-to-date and relevant data for the study.
• Pros:
  • Specific to the researcher’s requirements.
  • More control over the data quality.
• Cons:
  • Can be time-consuming and costly to gather.
  • Risk of bias in data collection methods.

3.7.1.2 Secondary Data

• Definition: Secondary data refers to data that was collected by someone else for a different purpose but is used by a researcher for their study.
• Sources: Includes government publications, websites, books, journal articles, internal records of organizations, and previously conducted studies.
• Examples:
  • Using census data for demographic studies.
  • Analyzing data from scientific journals for a literature review.
• Uses:
  • Useful for obtaining a broad understanding of the topic.
  • Helpful in comparing and corroborating primary data findings.
• Pros:
  • Less expensive and less time-consuming to collect.
  • Often covers a broader scope than primary data.
• Cons:
  • Might not be perfectly aligned with the current research needs.
  • Potential issues with relevance, accuracy, and timeliness.

3.7.2 All Types of Data

Here’s a comprehensive table of different types of data, in both English and Roman Urdu:

3.7.2.1 Table in English:

Data Type	Definition	Examples	Pros	Cons
Primary Data	Data collected directly by the researcher for their study. Original and source-based.	Surveys, Experiments, Observations	Tailored to specific needs, Control over data quality	Time-consuming and costly, Potential bias in collection
Secondary Data	Data collected by someone else for a different purpose but used by a researcher for their study.	Census data, Scientific journals, Organizational records	Less costly and time-consuming, Broader scope	May not align with current research needs, Relevance and timeliness issues
Quantitative Data	Numerical data that can be measured or counted.	Age, Temperature, Sales figures	Suitable for statistical analysis and making predictions	Requires numerical competence, May overlook contextual details
Qualitative Data	Descriptive data observed but not measured.	Colors, Text responses, Interview transcripts	Provides depth and context, Ideal for patterns and themes in data	Time-consuming to analyze, Subject to interpretational biases
Discrete Data	Numerical data with specific values and countable.	Number of students, Survey responses	Ideal for countable scenarios, Clear and distinct data points	Limited by its non-continuous nature
Continuous Data	Numerical data that can take any value within a range.	Height, Weight, Time	Allows for precise measurements, Suitable for scientific research	Can be complex to analyze, Requires sophisticated measurement tools
Categorical Data	Data grouped into categories.	Blood type, Brand names, Types of cuisine	Useful for classification and sorting, Easier to organize and interpret	Lacks numerical depth, Not suitable for mathematical analysis
Ordinal Data	Categorical data with a clear ordering.	Customer satisfaction ratings, Class ranks	Useful for ranking and ordering data, Provides more detail than nominal data	Intervals between ranks may not be equal
Nominal Data	Categorical data without a logical order.	Gender, Nationality, Marital status	Ideal for labeling or categorizing, Simple to organize	Lacks depth and order, Limited analytical use
Binary Data	Data with only two possible values.	Yes/No, True/False, On/Off	Simple and clear, Ideal for decision-making processes	Lacks complexity, Limited to two outcomes
Time-Series Data	Data points collected or recorded at regular time intervals.	Stock prices over time, Daily temperature readings	Ideal for trend analysis and forecasting	Can be complex to analyze, affected by time-related biases
Cross-Sectional Data	Data collected at a single point in time or over a very short period.	One-time surveys, Snapshot of sales data	Useful for capturing a specific moment, Easier to collect	Lacks longitudinal depth, May not capture changes over time
Longitudinal Data	Data collected over a long period to analyze changes.	Long-term health studies, Employee performance tracking	Ideal for observing changes over time, Provides depth and progression	Time-consuming to collect, Requires long-term commitment
Spatial Data	Data related to geographical or spatial locations.	GIS data, Navigation maps	Useful for geographical analysis, Supports mapping and spatial studies	Requires specialized tools and knowledge, Can be complex to interpret
Multidimensional Data	Data with multiple dimensions or aspects, often seen in complex databases.	Business intelligence data, Complex databases	Allows for deep data analysis, Ideal for business intelligence	Complex to analyze, Requires sophisticated tools and expertise
Unstructured Data	Data that doesn’t fit into a conventional database structure, like text, video, or audio data.	Videos, Audio recordings, Social media posts	Rich in information and context, Ideal for AI and machine learning applications	Challenging to organize and analyze, Requires advanced processing tools
Structured Data	Highly organized data that can easily be stored and queried in a database, like rows and columns in a spreadsheet.	Database records, Excel spreadsheets	Easy to access and manipulate, Ideal for traditional database management	May lack flexibility, Can be insufficient for complex data analysis
Semi-Structured Data	A blend of structured and unstructured data, like JSON or XML files.	JSON files, XML data	Balances flexibility and organization, Ideal for web data	Can be challenging to parse, Requires specific processing tools

---

3.7.2.2 Table in Roman Urdu:

Data Ki Qism	Tafseel	Misaalein	Fawaid	Nuqsanat
Primary Data	Data jo researcher ne apne study ke liye barah-e-raast jama kiya ho.	Surveys, Tajurbaat, Mushahedat	Khas zarurat ke mutabiq, Data ki quality par control	Waqt aur raqam ki zyada kharch, Collection mein bias ka imkan
Secondary Data	Data jo kisi aur ne dusre maqsad ke liye jama kiya ho lekin kisi researcher ne apne study ke liye istemal kiya ho.	Mardum shumari ke data, Scientific journals, Idarati records	Kam kharch aur waqt ki bachat, Zyada daire mein istemal	Mojudah research ki zaruraton se na milna, Mutabiqat aur waqt ki paabandi
Quantitative Data	Numerical data jo measure ya ginna ja sakta hai.	Umar, Darja-e-hararat, Sales figures	Statistical analysis aur predictions ke liye munasib	Numerical samajh ki zarurat, Context ko nazarandaz kar sakta hai
Qualitative Data	Tafseeli data jo dekha ja sakta hai lekin measure nahi kiya ja sakta.	Rang, Text jawabat, Interview transcripts	Gahraai aur context faraham karta hai, Patterns aur themes ko samajhne mein madadgar	Analyze karne mein waqt ka kharch, Tafseeri biases ka khadsha
Discrete Data	Numerical data jisme mukammal aur alag alag qiymat hoti hai.	Talba ki tadad, Survey ke jawabat	Ginne layak scenarios ke liye behtareen, Wazeh aur mukammal data points	Non-continuous hone ki wajah se hadbandi
Continuous Data	Numerical data jo kisi range mein kisi bhi qiymat le sakta hai.	Qad, Wazan, Waqt	Precise measurements ke liye munasib, Scientific research ke liye aham	Analyze karne mein pechida, Advanced measurement tools ki zarurat
Categorical Data	Data jo categories mein group kiya gaya ho.	Blood type, Brand names, Khaane ke iqsaam	Classification aur sorting mein aasan, Organize karne aur interpret karne mein sahulat	Numerical gehraai ki kami, Mathematical analysis ke liye na-munasi…
Ordinal Data	Categorical data jisme wazeh tarteeb ya sequence hoti hai.	Customer satisfaction ratings, Class ranks	Ranking aur ordering data ke liye munasib, Nominal data se zyada tafseelat faraham karta hai	Ranks ke darmiyan waqfa barabar na hona
Nominal Data	Categorical data jisme koi logical tarteeb ya order nahi hota.	Gender, Nationality, Marital status	Labeling ya categorizing ke liye ideal, Asaani se organize karna	Gehraai aur order ki kami, Analytical istemal mehdood
Binary Data	Sirf do mumkinah qiymaton wala data.	Yes/No, True/False, On/Off	Saada aur wazeh, Faisla-sazi ke amal ke liye munasib	Pechida maaloomat ki kami, Sirf do outcomes tak mehdood
Time-Series Data	Data points jo baqaida waqt ke faslay par jama kiye gaye ho.	Stock prices waqt ke sath, Rozana darja-e-hararat ki readings	Rujhanat aur forecasting ke liye munasib	Analyze karne mein pechida, Waqt se mutaliq biases se mutasir
Cross-Sectional Data	Data jo aik specific waqt mein ya bohot hi mukhtasir muddat mein jama kiya gaya ho.	Aik waqt mein kiye gaye surveys, Sales data ka snapshot	Kisi khaas lamhe ko capture karne ke liye munasib, Jama karne mein asaan	Longitudinal gehraai ki kami, Waqt ke sath tabdeeliyon ko capture nahi karta
Longitudinal Data	Data jo lambay waqt ke douran jama kiya gaya ho tabdeeliyon ka analysis karne ke liye.	Long-term sehat ki studies, Employee performance tracking	Waqt ke sath tabdeeliyon ko dekhne ke liye munasib, Gehraai aur taraqqi faraham karta hai	Jama karne mein waqt ka kharch, Long-term commitment ki zarurat
Spatial Data	Geographical ya spatial locations se mutaliq data.	GIS data, Navigation maps	Geographical analysis ke liye munasib, Mapping aur spatial studies mein madadgar	Khaas tools aur ilm ki zarurat, Interpret karne mein pechida
Multidimensional Data	Kai dimensions ya aspects wala data, jo aksar pechida databases mein hota hai.	Business intelligence data, Pechida databases	Deep data analysis ke liye munasib, Business intelligence ke liye aham	Analyze karne mein pechida, Advanced tools aur expertise ki zarurat
Unstructured Data	Wo data jo riwayati database structure mein fit nahi hota, jaise text, video, ya audio data.	Videos, Audio recordings, Social media posts	Maloomat aur context mein maaloomat se bharpoor, AI aur machine learning applications ke liye ideal	Organize aur analyze karne mein challenging, Advanced processing tools ki zarurat
Structured Data	Intehai tarteeb shuda data jo asaani se database mein store aur query kiya ja sakta hai, jaise spreadsheet mein rows aur columns.	Database records, Excel spreadsheets	Access aur manipulate karne mein aasan, Riwayati database management ke liye ideal	Lachak ki kami, Complex data analysis ke liye na-kaafi
Semi-Structured Data	Structured aur unstructured data ka mixture, jaise JSON ya XML files.	JSON files, XML data	Lachak aur tarteeb ka balance, Web data ke liye ideal	Parse karne mein challenging, Khaas processing tools ki zarurat

---

3.8 Follow us

Follow us

Main umeed karta hun k ap ko ye chapter ne bht kuch seekhaya ho ga, or agar sach main seekhaya hy then please do support us by sharing this book with your friends and colleagues. Also, do share your feedback with us, so that we can improve our work in future.

---