Python मध्ये EDA — नवशिक्यांची checklist
Read this guide in: English · मराठी · हिंदी
EDA म्हणजे प्रश्नांसोबत table explore करणं — random stare नाही. pandas checklist: shape, head, info, missing, साधा groupby, मग findings लिहा. Infosys-style hours: client A = 70.
चला मित्रांनो! CSV उघडणं म्हणजे EDA नाही. EDA = प्रश्न → inspect → note. आज shop, Instagram-like, Netflix-style किंवा client sheets वर reuse होणारी first-pass checklist. समजलं का?
थोडक्यात उत्तर
उभा first-pass EDA:
- एक business प्रश्न लिहा.
read_csv→shape→head→columns.info()+isna().sum().- Numbers:
describe(); categories:value_counts. - प्रश्न उत्तरणारा एक groupby.
- Findings bullets + numbers (chart पुढच्या guide मध्ये).
question → inspect → clean notes → groupby → findings
Language: English · मराठी · हिंदी
या guide आधी काय हवं?
- pandas load / filter / groupby / missing basics.
- Course: First-pass checklist.
Before आणि after
Before — Infosys-style client hours dump; no questions.
After — EDA checklist; top client A = 70 hours.
Before: raw hours dump. After: checklist भरली; top client A = 70.
EDA काय आहे (आणि काय नाही)
आहे
- प्रश्नाने guided structured looking.
- Data-quality issues लवकर सापडणं.
- Number-backed छोटी findings.
या course वर नाही
- sklearn models train नाही.
- वाक्य नसलेला ४०-chart firework नाही.
- Cleaning कायम skip नाही.
Infosys-style sample
| client | hours |
|---|---|
| A | 40 |
| B | NaN |
| A | 30 |
| C | 20 |
प्रश्न: कोणत्या client कडे सर्वाधिक recorded hours?
उभा pass:
- shape →
(4, 2) - missing → hours मध्ये 1 NaN
- fill/drop decide — document
- Known hours: A 70, C 20, B unknown
- Finding: “या extract वर client A recorded hours मध्ये 70 ने पुढे.”
Checklist (दृश्यमान ठेवा)
- Question — एक वाक्य.
- Load — path, sheet, nrows.
- Head / columns
- Grain — एक row म्हणजे काय?
- Types
- Missing
- Duplicates
- Univariate — describe / value_counts
- Relationships — groupby / crosstab
- Findings — numbers असलेले bullets
- Next chart
Crosstab आणि correlation caution
Correlation ≠ causation. Course: Crosstab and correlation caution.
Findings document
वाईट: “Data interesting वाटतं.”
चांगलं:
- “Online shop sales total 190 (Mumbai 120, Pune 70).”
- “एका Netflix-style title ला missing watches; count KPIs साठी 0 fill.”
- “Client A = 70 hours; B source मध्ये incomplete.”
Quick drills
- Shop: channel जिंकतो? → Online 190.
- Engagement: city likes? → Mumbai 120 vs Pune 70.
- Checklist तीच — फक्त प्रश्न बदलतो.
Timebox
छोट्या file वर first pass: एका तासाखाली. Depth = तेजस्वी प्रश्न, इंद्रधनुष्य charts नाही.
Ravindra Bagale's Tip
Notebook च्या top Markdown cell मध्ये प्रश्न ठेवा. Cell त्या प्रश्नाला सेवा देत नसेल तर खाली हलवा किंवा काढा. तेच professional EDA. समजलं का?
Practice task
चार-row client hours:
- प्रश्न एका वाक्यात.
- shape, head, isna().
- non-missing वर groupby client sum.
- तीन finding bullets — A = 70 सहित.
- पुढचा chart नाव (bar of hours by client).
Learn it properly
Related: Handle missing values · First charts
समजलं का? EDA = प्रश्न-driven pandas + लिहिलेल्या findings — ML नको. Client A 70 = finding, vibe नाही. आता पुढे जाऊया.
Frequently asked questions
What is EDA?
Exploratory data analysis: structured looking before you claim insights or build dashboards.
Is EDA the same as cleaning?
Cleaning fixes types/missing/dupes. EDA uses clean-enough data to answer questions. They overlap in real notebooks.
Do I need seaborn for EDA?
Tables first. Charts help storytelling — covered in the visualisation guides/course chapters.
How long should first-pass EDA take?
For a small CSV, under an hour for a clear checklist. Depth comes from better questions.
Correlation caution?
High correlation is not proof of cause. The course has a caution lesson — read it.
Course links?
EDA chapter: question-driven EDA and first-pass checklist.