Data Analytics Course in Chennai

Live Demo: What You'll Learn to Do

Click through each step below. See real data transform into actionable business insights. This is what 6 months of training gets you.

1. Data Wrangling: Raw → Clean → Insights

1Messy Data

2Cleaned

3Pivot Analysis

4Dashboard

↔ Click any step above to jump back and forth freely.

💡Why clean data first? Real business data is always messy — duplicates, typos and mixed formats silently corrupt every report built on top. Analysts spend ~80% of their time here because wrong data means wrong ₹-decisions.

⚠️ 6 Data Quality Issues Detected (500 rows scanned):

47 duplicate rows — same order counted multiple times
Mixed date formats — DD/MM/YYYY, YYYY-MM-DD, MM/DD/YYYY all present
Mixed currency — ₹ and $ values in the same column
12 null revenues — blank cells that break SUM()
Trailing spaces — "John Smith " ≠ "John Smith" to a computer
Inconsistent case — PAID / paid / Paid treated as 3 statuses

📄 raw_sales_export.csv — untouched database dump (showing 18 of 500 rows)

Customer	Order Date	Product	Region	Revenue	Status
John Smith	01/15/2024	Laptop	North	₹ 85,000	PAID
Sarah Lee	2024-01-15	Phone	South	$550	paid
John Smith	01/15/2024	Laptop	North	₹ 85,000	PAID
Mike Johnson	2024/01/16	(blank)	East	₹ 45,000	Paid
Sarah Lee	15-01-2024	Phone	South	(null)	PENDING
Priya Sharma	2024-01-17	Tablet	West	₹ 28,000	paid
John Smith	01/15/2024	Laptop	North	₹ 85,000	PAID
Raj Patel	2024-01-18	Monitor	North	₹ 35,000	PAID
Mike Johnson	18/01/2024	Keyboard	East	₹ 12,000	paid
Sarah Lee	2024-01-19	Phone	South	₹ 38,000	PENDING
Priya Sharma	2024-01-20	Tablet	West	₹ 28,000	paid
Akshay Kumar	20/01/2024	Laptop	East	(null)	PAID
Deepa Singh	2024-01-21	Phone	North	₹ 42,000	paid
Raj Patel	2024-01-22	Monitor	North	₹ 35,000	PAID
Anita Verma	2024-01-24	Laptop	South	₹ 92,000	paid
Vikram Rao	2024-01-25	Monitor	East	₹ 38,000	PENDING
Neha Gupta	25/01/2024	Phone	West	₹ 42,000	PAID
John Smith	01/15/2024	Laptop	North	₹ 85,000	PAID

💡What changed? Same data, now trustworthy. Compare the two views below — duplicates gone, one date format, one currency, nulls filled, statuses standardized. Every number downstream now ties out.

Before — raw

Customer	Date	Revenue	Status
John Smith	01/15/2024	₹ 85,000	PAID
John Smith	01/15/2024	₹ 85,000	PAID
Sarah Lee	15-01-2024	$550	paid
Mike Johnson	2024/01/16	₹ 45,000	Paid
Sarah Lee	15-01-2024	(null)	PENDING
Akshay Kumar	20/01/2024	(null)	PAID
Priya Sharma	2024-01-17	₹ 28,000	paid

After — cleaned

Customer	Date	Revenue (₹)	Status
John Smith	2024-01-15	85,000	PAID
Sarah Lee	2024-01-15	45,650	PAID
Mike Johnson	2024-01-16	45,000	PAID
Sarah Lee	2024-01-15	38,000	PENDING
Akshay Kumar	2024-01-20	52,400	PAID
Priya Sharma	2024-01-17	28,000	PAID
Raj Patel	2024-01-18	35,000	PAID

✓ 7 Transformations Applied Automatically:

Removed 47 duplicate rows (500 → 453 clean rows)
Standardized all dates to ISO YYYY-MM-DD
Converted $ to ₹ at ₹83/$ and stripped symbols → pure numbers
Imputed 12 null revenues using product-region median
Trimmed whitespace and title-cased all names
Normalized status values to UPPERCASE
Validated — data quality score 98.5%

Clean Rows

453

from 500 raw

Issues Fixed

127

100% resolved

Duplicates Killed

₹39.9L de-inflated

Quality Score

98.5%

analysis-ready

💡Why pivot? 453 rows are unreadable. A pivot collapses them into a cross-tab that answers "which product sells where, and how much does each contribute?" — the colour heatmap shows strength at a glance.

📊 Pivot 1 — Revenue by Product × Region (heatmap + share)

Product ↓ / Region →	North	South	East	West	Total	% Share
Laptop	₹1,78,000	₹1,92,000	₹92,000	₹85,000	₹5,47,000	50.3%
Phone	₹42,000	₹80,000	₹38,000	₹42,000	₹2,02,000	18.6%
Monitor	₹73,000	₹35,000	₹38,000	₹22,000	₹1,68,000	15.5%
Tablet	₹28,000	₹18,000	₹14,000	₹56,000	₹1,16,000	10.7%
Keyboard	₹15,000	₹12,000	₹18,000	₹9,000	₹54,000	5.0%
Grand Total	₹3,36,000	₹3,37,000	₹2,00,000	₹2,14,000	₹10,87,000	100%

📈 Pivot 2 — Product performance with calculated fields

Product	Units Sold	Revenue	Avg Order Value	Margin %	QoQ Growth
Laptop	89	₹5,47,000	₹61,460	22%	+18% ↑
Phone	156	₹2,02,000	₹12,950	35%	+9% ↑
Monitor	78	₹1,68,000	₹21,540	28%	+5% ↑
Tablet	64	₹1,16,000	₹18,125	31%	−3% ↓
Keyboard	112	₹54,000	₹4,820	42%	+12% ↑

Revenue Leader

Laptop

50.3% of total

Top Region

South

₹3,37,000

Highest Margin

Keyboard

42% margin

⚠ Watch

Tablet

−3% declining

🎯This is the payoff. From a messy CSV to a live executive dashboard — KPIs, trends and the product mix on one screen. This exact skill (raw data → decisions) is what companies pay analysts ₹5L-₹12L to do.

📊 Sales Performance Dashboard

Sep 2023 – Feb 2024 · Auto-refreshed from cleaned dataset

● LIVE

Total Revenue

₹10,87,000

↑ 14% MoM

Total Orders

499

↑ 8% MoM

Avg Order Value

₹21,783

↑ 6% MoM

Repeat Rate

38%

↑ 4 pts

Top Region

South

₹3,37,000 · 31%

Customer	Orders	Revenue	Avg Order	Segment
John Smith	8	₹3,15,000	₹39,375	VIP
Anita Verma	5	₹1,84,000	₹36,800	VIP
Akshay Kumar	4	₹1,52,000	₹38,000	Regular
Deepa Singh	4	₹98,000	₹24,500	Regular
Raj Patel	3	₹87,000	₹29,000	Regular

2. SQL: Querying Data at Real Scale

1Raw Database

2Basic Query

3Joins + Windows

4Forecasting

↔ Click any step above to explore in any order.

💡Why SQL over Excel? Excel dies past ~1M rows. Real databases hold millions across many linked tables. SQL is how you ask questions of data at that scale — it's the #1 skill in every analyst job posting.

The data lives in 3 separate tables connected by IDs. Nothing is readable yet — you have to join and aggregate.

sales (500K rows)

id	cust_id	prod	amt
5001	C123	P01	85000
5002	C456	P02	38000
5003	C123	P03	45000
5004	C789	P01	92000
5005	C456	P04	28000
5006	C123	P02	42000

customers (174 rows)

id	name	region
C123	John Smith	North
C456	Sarah Lee	South
C789	Anita Verma	South
C246	Raj Patel	East
C802	Neha Gupta	West
C913	Akshay K.	East

products (5 rows)

id	name	cat
P01	Laptop	Compute
P02	Phone	Mobile
P03	Monitor	Display
P04	Tablet	Mobile
P05	Keyboard	Accessory

💡Start simple. A GROUP BY collapses 500,000 sales rows into one line per customer — total spent, order count, averages — in a fraction of a second.

SELECT
  customer_id,
  COUNT(*)              AS orders,
  SUM(amount)           AS total_spent,
  ROUND(AVG(amount))    AS avg_order,
  MAX(order_date)       AS last_order
FROM sales
WHERE order_date >= '2024-01-01'
GROUP BY customer_id
ORDER BY total_spent DESC;

Rows Scanned

500K

full table

Rows Returned

174

1 per customer

Exec Time

0.08s

try that in Excel

Result set — customer spend summary

customer_id	orders	total_spent	avg_order	last_order
C123	8	₹3,15,000	₹39,375	2024-02-14
C789	5	₹1,84,000	₹36,800	2024-02-11
C913	4	₹1,52,000	₹38,000	2024-02-09
C456	4	₹98,000	₹24,500	2024-02-08
C246	3	₹87,000	₹29,000	2024-02-05
C802	2	₹42,000	₹21,000	2024-01-28

💡Now the real work. JOIN stitches the 3 tables into readable names, and window functions add per-customer rankings, running totals and share-of-wallet — the analysis hiring managers test for.

SELECT
  c.name, c.region, p.name AS product, s.amount,
  RANK()       OVER (PARTITION BY c.id ORDER BY s.amount DESC)          AS rank_in_cust,
  SUM(s.amount) OVER (PARTITION BY c.id ORDER BY s.order_date)          AS running_total,
  ROUND(100.0 * s.amount /
        SUM(s.amount) OVER (PARTITION BY c.id), 1)                      AS pct_of_wallet
FROM sales s
JOIN customers c ON s.customer_id = c.id
JOIN products  p ON s.product_id  = p.id
ORDER BY c.name, s.amount DESC;

Tables Joined

sales+cust+prod

Window Funcs

RANK, SUM, %

Exec Time

0.34s

500K rows

Enriched result — names, rankings, running totals, share of wallet

Customer	Region	Product	Amount	Rank	Running Total	% Wallet
John Smith	North	Laptop	₹1,10,000	1	₹1,10,000	34.9%
John Smith	North	Laptop	₹85,000	2	₹1,95,000	27.0%
John Smith	North	Monitor	₹78,000	3	₹2,73,000	24.8%
John Smith	North	Phone	₹42,000	4	₹3,15,000	13.3%
Anita Verma	South	Laptop	₹92,000	1	₹92,000	50.0%
Anita Verma	South	Tablet	₹56,000	2	₹1,48,000	30.4%
Anita Verma	South	Phone	₹36,000	3	₹1,84,000	19.6%

🎯Analyst-level SQL. CTEs + LAG() compare each month to the last, compute growth, and project next month — forecasting straight in the database. This is exactly what you'll do on day one of the job.

WITH monthly AS (
  SELECT DATE_TRUNC('month', s.order_date) AS month, c.region,
         SUM(s.amount) AS revenue
  FROM sales s JOIN customers c ON s.customer_id = c.id
  GROUP BY 1, 2
),
growth AS (
  SELECT month, region, revenue,
         LAG(revenue) OVER (PARTITION BY region ORDER BY month) AS prev,
         ROUND(revenue * 1.18) AS forecast_next
  FROM monthly
)
SELECT month, region, revenue, prev,
       ROUND(100.0*(revenue-prev)/prev, 1) AS growth_pct, forecast_next
FROM growth WHERE prev IS NOT NULL
ORDER BY growth_pct DESC;

Regional growth + next-month forecast

Month	Region	Revenue	Prev Month	Growth %	Forecast (next)
Feb 2024	South	₹3,37,000	₹2,84,000	+18.7%	₹3,97,660
Feb 2024	North	₹3,36,000	₹2,89,000	+16.3%	₹3,96,480
Feb 2024	West	₹2,14,000	₹1,96,000	+9.2%	₹2,52,520
Feb 2024	East	₹2,00,000	₹1,88,000	+6.4%	₹2,36,000

Regional revenue — last 3 months + forecast

💡 What the query reveals:
• South and North are neck-and-neck and both growing ~17% — the battleground for budget.
• Every region is positive; forecast projects ₹12.8L next month (+18%).
• East lags at +6.4% — lowest penetration, biggest untapped upside.

3. Pattern Recognition & Customer Segmentation

1Raw Behaviour

2Find Patterns

3RFM Segments

4Forecast + Scenarios

↔ Jump between steps anytime.

💡Why look for patterns? Buried in a flat transaction log are hidden groups — loyal spenders, one-timers, people about to churn. Spotting them is how businesses stop blanket-marketing and start targeting.

Raw behaviour log — 453 transactions, 174 customers (sample)

Customer	Last Purchase	Total Spent	Orders (6mo)	Days Since Last	Avg Gap (days)
C123	2024-02-14	₹3,15,000	8	2	7
C789	2024-02-11	₹1,84,000	5	5	11
C456	2024-02-08	₹98,000	4	8	14
C246	2024-01-18	₹42,000	2	34	45
C551	2023-12-20	₹15,000	1	68	—
C802	2024-02-06	₹42,000	2	10	22
C913	2024-02-09	₹1,52,000	4	7	13
C377	2024-01-05	₹28,000	1	42	—

💡RFM = Recency, Frequency, Monetary. Scoring every customer on how recently, how often and how much they buy sorts all 174 into actionable segments — so you know exactly who to reward and who to rescue.

Champions (VIP)

42% of revenue

Loyal

35% of revenue

Potential

12% of revenue

At-Risk

8% · slipping

Dormant

3% · 60+ days

Segment scorecard with recommended actions

Segment	Customers	Avg LTV	Frequency	Recency	% Revenue	Risk	Recommended Action
Champions	18	₹3,50,000	8.2/mo	2 days	42%	Low	Loyalty perks, early access, upsell
Loyal	64	₹85,000	3.5/mo	6 days	35%	Low	Cross-sell bundles, referrals
Potential	40	₹45,000	2.1/mo	12 days	12%	Med	Nurture emails, first-repeat nudge
At-Risk	28	₹42,000	1.2/mo	34 days	8%	Med	Re-engagement offer, check-in
Dormant	34	₹15,000	0.3/mo	68 days	3%	High	Aggressive win-back discount

Avg lifetime value by segment

Share of revenue by segment

🎯From insight to ₹. Now model what happens if you act. The forecast projects baseline revenue; the scenario table shows how much each retention play adds — and its ROI. This is analytics that pays for itself.

This Month

₹10,87,000

actual (Feb)

Next Month

₹11,24,000

baseline +3.4%

Q2 Estimate

₹35,80,000

±₹1.8L

Confidence

87%

backtested

6-month revenue — actual vs forecast (with confidence band)

Scenario analysis — what each retention play is worth

Scenario	Investment	Jun Revenue	vs Baseline	ROI	Feasibility
Baseline (do nothing)	₹0	₹11,24,000	—	—	—
Win-back 30% Dormant	₹34,000	₹12,54,000	+₹1,30,000	3.8x	Medium
Upsell VIPs	₹90,000	₹12,86,000	+₹1,62,000	1.8x	High
Retain 50% At-Risk	₹56,000	₹11,92,000	+₹68,000	1.2x	High
Combined strategy	₹1,80,000	₹13,98,000	+₹2,74,000	1.5x	Medium

📈 The recommendation writes itself:
• Win-back on Dormant customers is the highest-ROI play at 3.8x — start there.
• Combined strategy adds ₹2.74L/month for ₹1.8L spend — a ₹11L+ annual swing.
• 42% of revenue sits in 18 Champions — a retention program here is non-negotiable.

4. Python: Automate the Entire Pipeline

1The Script

2Execute

3Results

4Automation ROI

↔ Steps are freely navigable — revisit the code or results anytime.

💡Why Python? Everything in demos 1–3 — cleaning, joining, segmenting, forecasting — collapses into one script that runs in seconds on any dataset size. Write once, run forever. This is "automation" on your resume.

import pandas as pd, numpy as np
from sklearn.preprocessing import StandardScaler
from sklearn.cluster import KMeans
from statsmodels.tsa.arima.model import ARIMA

# 1. Load + clean  (replaces Demo 1)
df = pd.read_csv('sales.csv').drop_duplicates()
df['amount'] = df['amount'].fillna(df.groupby('product')['amount'].transform('median'))
df['date']   = pd.to_datetime(df['date'], format='mixed')

# 2. Join reference tables  (replaces Demo 2)
df = df.merge(customers, on='customer_id').merge(products, on='product_id')

# 3. RFM segmentation  (replaces Demo 3)
snapshot = df['date'].max() + pd.Timedelta(days=1)
rfm = df.groupby('customer_id').agg(
    recency   = ('date',   lambda x: (snapshot - x.max()).days),
    frequency = ('date',   'count'),
    monetary  = ('amount', 'sum'))
rfm_scaled = StandardScaler().fit_transform(rfm)
rfm['segment'] = KMeans(n_clusters=5, n_init=10).fit_predict(rfm_scaled)

# 4. Forecast next month
monthly = df.set_index('date').resample('M')['amount'].sum()
forecast = ARIMA(monthly, order=(1,1,1)).fit().forecast(1)

rfm.to_csv('segments.csv'); monthly.to_csv('monthly.csv')
print(f"✓ Done. Forecast: ₹{forecast.iloc[0]:,.0f}")

🎯The whole pipeline — automated. A full day of analyst work now finishes before your coffee. Clean data, 5 segments and a forecast, generated with zero manual steps.

Runtime

4.2s

vs ~6 hrs manual

Rows Processed

500K

100% cleaned

Forecast Accuracy

89%

backtested

Files Output

CSV + charts

Monthly revenue — 6 months actual + ARIMA forecast

Auto-generated segments.csv (KMeans, 5 clusters)

Segment	Customers	Avg Recency	Avg Frequency	Avg Monetary	Auto-Label
Cluster 0	18	2 days	8.2	₹3,50,000	Champions
Cluster 1	64	6 days	3.5	₹85,000	Loyal
Cluster 2	40	12 days	2.1	₹45,000	Potential
Cluster 3	28	34 days	1.2	₹42,000	At-Risk
Cluster 4	34	68 days	0.3	₹15,000	Dormant

⏱ Time per report	~6 hours
🔁 Frequency feasible	Weekly (max)
📉 Max data size	~1M rows
❌ Error rate	Human, frequent
💰 Analyst cost/yr	High (all manual)

⏱ Time per report	4.2 seconds
🔁 Frequency feasible	Every hour, scheduled
📈 Max data size	Billions of rows
✅ Error rate	Zero, reproducible
💰 Value delivered	10x analyst output

Turn Data Into Decisions. Master Analytics in 6 Months.

Join Next Batch

The Data Advantage: Will You Choose Wisely?

Real Scenario From Our Students' Workplaces:

Live Demo: What You'll Learn to Do

1. Data Wrangling: Raw → Clean → Insights

⚠️ 6 Data Quality Issues Detected (500 rows scanned):

✓ 7 Transformations Applied Automatically:

📊 Sales Performance Dashboard

2. SQL: Querying Data at Real Scale

3. Pattern Recognition & Customer Segmentation

✓ 5 Behavioural Patterns Detected:

4. Python: Automate the Entire Pipeline

What Our Students Say

Rajesh Kumar

Priya Sharma

Arjun Patel

Deepa Iyer

Your 6-Month Learning Path

By End of Course, You'll Be Able To:

Transparent Pricing

⏰ Limited Seats Available

Frequently Asked Questions

Do I need coding experience?

How long until I get a job?

What if I can't keep up?

Is there a money-back guarantee?

Ready to Make Data-Driven Decisions?