Advanced Tableau

🎯 Mục tiêu bài học

TB5 min

Sau bài học này, bạn sẽ:

✅ Master LOD Expressions: Cohort, CLV, New vs Returning

✅ Analytics pane: Trend Line, Forecast, Clustering

✅ Tableau Prep basics cho data cleaning

✅ Performance optimization: Extracts, Context Filters

✅ Advanced: Set Actions, Dynamic Zone, Small Multiples

Thời gian: 45 phút | Độ khó: Advanced | Tool: Tableau Desktop

Task 0

📖 Bảng Thuật Ngữ Quan Trọng

TB5 min

Thuật ngữ	Tiếng Việt	Mô tả
Cohort Analysis	Phân tích đoàn hệ	Group customers by acquisition period
CLV	Giá trị vòng đời KH	Customer Lifetime Value
Trend Line	Đường xu hướng	Linear, polynomial, exponential fit
Forecast	Dự báo	Time-series prediction
Clustering	Phân cụm	K-means grouping tự động
Tableau Prep	Chuẩn bị dữ liệu	Visual ETL tool
Extract (.hyper)	Trích xuất	Local data cache cho performance
Context Filter	Bộ lọc ngữ cảnh	Filter ưu tiên, áp dụng trước
Set Action	Hành động tập hợp	User select → update set → calcs react
Small Multiples	Nhiều biểu đồ nhỏ	Same chart repeated per category

Checkpoint

LOD = FIXED/INCLUDE/EXCLUDE. Cohort = group by acquisition. Extract (.hyper) = local cache. Context Filter = priority filter. Bạn nhớ 3 loại LOD chưa?

Task 1

🎯 1. LOD Expressions Deep Dive

TB5 min

1.1 Cohort Analysis

Ví dụ

1// Customer Acquisition Year
2{ FIXED [Customer ID] : MIN(YEAR([Order Date])) }
3 
4// Use: Color by cohort year → see retention over time

1.2 Customer Lifetime Value

Ví dụ

1// Total revenue per customer
2{ FIXED [Customer ID] : SUM([Sales]) }
3 
4// Order frequency
5{ FIXED [Customer ID] : COUNTD([Order ID]) }
6 
7// First & Last order
8{ FIXED [Customer ID] : MIN([Order Date]) }
9{ FIXED [Customer ID] : MAX([Order Date]) }
10 
11// Active days
12DATEDIFF('day',
13  { FIXED [Customer ID] : MIN([Order Date]) },
14  { FIXED [Customer ID] : MAX([Order Date]) }
15)

1.3 New vs Returning Customers

Ví dụ

1IF [Order Date] = { FIXED [Customer ID] : MIN([Order Date]) }
2THEN "New Customer"
3ELSE "Returning Customer"
4END

1.4 Percent of Parent

Ví dụ

1// Sub-category as % of category
2SUM([Sales]) / { FIXED [Category] : SUM([Sales]) }
3 
4// Product as % of total
5SUM([Sales]) / { FIXED : SUM([Sales]) }

LOD Use Cases thực tế: Customer cohort analysis, CLV calculation, New vs Returning segmentation, % of parent. FIXED là powerful nhất — bỏ qua mọi viz filters!

Checkpoint

Cohort = FIXED Customer : MIN(YEAR(Date)). CLV = FIXED Customer : SUM(Sales). New vs Returning = compare Order Date với FIXED MIN. Mỗi LOD giải quyết 1 business question!

Task 2

📊 2. Analytics Pane

TB5 min

2.1 Available Analytics

📊Analytics Pane

📏Summarize: Average, Median, Constant Line

🔮Model: Trend Line, Forecast, Cluster

⚙️Custom: Reference Line/Band, Distribution

2.2 Trend Line

Ví dụ

11. Analytics pane → Drag "Trend Line" onto chart
22. Models: Linear, Logarithmic, Polynomial, Power
33. Options: confidence bands, R-squared

2.3 Forecast

Ví dụ

11. Drag "Forecast" onto line chart
22. Options:
3   - Length: 3, 6, 12 months
4   - Prediction interval: 90%, 95%
5   - Seasonal pattern: Auto-detect

Best for: Sales forecasting, demand planning.

2.4 Clustering

Ví dụ

11. Drag "Cluster" onto scatter plot
22. Number of clusters: Auto or custom (2-10)
33. Result: points colored by cluster

Best for: Customer segmentation, product grouping.

Forecast trong Tableau dùng exponential smoothing — tự detect seasonal patterns. Rất tốt cho quick prediction nhưng không thay thế proper ML models!

Checkpoint

Analytics Pane: Trend Line (regression), Forecast (time-series), Cluster (k-means). Drag-and-drop lên chart — không cần code! Forecast dùng exponential smoothing.

Task 3

🔧 3. Tableau Prep

TB5 min

3.1 Why Tableau Prep?

Before Prep	With Prep
Manual Excel cleanup	Visual data cleaning
SQL transformations	Drag-drop operations
Data lives in silos	Union/join multiple sources
Repetitive process	Saved & scheduled flows

3.2 Key Operations

Operation	Use Case
Clean	Rename, remove, change types
Filter	Remove rows (null, invalid)
Group	Merge similar values (typos)
Pivot	Wide → long format
Union	Combine same-structure tables
Join	Combine different tables on key
Aggregate	Group by + summarize
Script	Python/R for advanced transform

3.3 Typical Prep Flow

Ví dụ

1Input (Excel files)
2    ↓
3Clean: Fix column names, data types
4    ↓
5Filter: Remove nulls, test data
6    ↓
7Join: Customer table + Orders table
8    ↓
9Calculate: Add profit margin, segments
10    ↓
11Output: Tableau Extract (.hyper)

Checkpoint

Tableau Prep = visual ETL tool. Flow: Input → Clean → Filter → Join → Calculate → Output (.hyper). Saved flows có thể schedule refresh tự động!

Task 4

⚡ 4. Performance Optimization

TB5 min

5.1 Performance Recording

Ví dụ

1Help → Settings → Start Performance Recording
2→ Navigate dashboard → Stop Recording → View bottlenecks

5.2 Optimization Tips

Area	Tip
Data	Use Extracts (.hyper) instead of Live
Filters	Context filters first, then normal
Calculations	Row-level calcs faster than LOD in some cases
Dashboard	Max 6-8 sheets per dashboard
Dates	Discrete dates, not continuous
Marks	Reduce number (< 10K ideal)

5.3 Extract Optimization

Ví dụ

1Data Source → Extract:
21. Add filters to reduce data size
32. Aggregate to visible dimensions
43. Hide unused fields
54. Schedule incremental refresh

Performance killers: Live connections to slow databases, too many marks (>100K), complex LOD stacking, excessive filters. Always use Extracts cho production dashboards!

Checkpoint

Extracts > Live connections. Context Filters áp dụng trước normal filters. Max 6-8 sheets per dashboard. < 10K marks ideal. Performance Recording để debug!

Task 5

🚀 5. Advanced Techniques

TB5 min

5.1 Set Actions

Ví dụ

1Dashboard → Actions → Change Set Values
2Set: [My Set]
3Run on: Select
4Clearing: Remove from set

Use case: User selects items → set updates → calculations react.

5.2 Dynamic Zone Visibility

Ví dụ

11. Create boolean calculated field
22. Control container visibility based on field value
3→ Dynamic dashboard layout!

5.3 Small Multiples

Ví dụ

11. Drag [Category] → Columns
22. Drag [Month] → Columns (nested)
33. Drag [Sales] → Rows
4→ One chart per category, same scale

Checkpoint

Set Actions = interactive set membership. Dynamic Zone = show/hide containers based on logic. Small Multiples = same chart repeated per category — great for comparison!

Task 6

📋 Tổng kết

TB5 min

Kiến thức đã học

Chủ đề	Nội dung chính	Tầm quan trọng
LOD Deep Dive	Cohort, CLV, New/Returning, % Parent	Advanced analysis
Analytics	Trend, Forecast, Cluster (built-in)	Predictive
Tableau Prep	Visual ETL, flows, scheduling	Data prep
Performance	Extracts, Context Filters, mark limits	Production
Advanced	Set Actions, Dynamic Zone, Small Multiples	Premium features

Câu hỏi tự kiểm tra

LOD Expression giải quyết được bài toán gì mà Table Calcs không làm được?
Tableau Prep khác Power Query thế nào?
Context Filter giúp cải thiện performance thế nào?
Set Actions dùng trong trường hợp nào?

Bài tiếp theo: DAX Fundamentals — Chuyển sang Power BI và học DAX!

🎉 Tuyệt vời! Bạn đã master Tableau nâng cao!

Nhớ: LOD + Analytics + Performance = Tableau Pro. Bạn đã sẵn sàng cho bất kỳ dự án Tableau nào!

Task 7

🎯 Mục tiêu bài học

📖 Bảng Thuật Ngữ Quan Trọng

Checkpoint

🎯 1. LOD Expressions Deep Dive

1.1 Cohort Analysis

1.2 Customer Lifetime Value

1.3 New vs Returning Customers

1.4 Percent of Parent

Checkpoint

📊 2. Analytics Pane

2.1 Available Analytics

2.2 Trend Line

2.3 Forecast

2.4 Clustering

Checkpoint

🔧 3. Tableau Prep

3.1 Why Tableau Prep?

3.2 Key Operations

3.3 Typical Prep Flow

Checkpoint

⚡ 4. Performance Optimization

5.1 Performance Recording

5.2 Optimization Tips

5.3 Extract Optimization

Checkpoint

🚀 5. Advanced Techniques

5.1 Set Actions

5.2 Dynamic Zone Visibility

5.3 Small Multiples

Checkpoint

📋 Tổng kết

Kiến thức đã học

Câu hỏi tự kiểm tra

Khóa học

Mentor & Hỗ trợ

Blog

Giới thiệu