Filters, Parameters, Sets, Groups and Hierarchies in Tableau

Filters

Show Filter (right-click the pill → Show Filter) puts an interactive control on the sheet: single value list, dropdown, multiple values, slider, wildcard match. Apply to Worksheets → All Using This Data Source / Selected Worksheets makes one filter drive several sheets on a dashboard.

import pandas as pd, numpy as np
df = pd.read_csv("superstore_sales.csv", parse_dates=["order_date"])
loss = df.groupby("sub_category").profit.sum().sort_values()
print("Loss-making sub-categories (measure filter SUM(Profit) < 0):")
print((loss[loss < 0] / 1e5).round(1))

Loss-making sub-categories (measure filter SUM(Profit) < 0):
sub_category
Tables      -8.2
Bookcases   -4.2
Name: profit, dtype: float64

Top N and context

Top N is computed after context filters but before ordinary dimension filters. So "Top 5 products in the South" needs the Region filter in context; otherwise Tableau finds the national Top 5 first and then keeps only those that happen to be in the South.

top_all = df.groupby("product_name").sales.sum().nlargest(5)
south = df[df.region == "South"]
top_south = south.groupby("product_name").sales.sum().nlargest(5)
print("National top 5:", list(top_all.index))
print("South top 5 (Region in context):", list(top_south.index))
print("Overlap:", len(set(top_all.index) & set(top_south.index)), "of 5")

National top 5: ['Orbit Copier 746', 'Apex Copier 347', 'Vertex Copier 262', 'Zenith Copier 719', 'Zenith Machine 401']
South top 5 (Region in context): ['Orbit Copier 746', 'Vertex Copier 262', 'Zenith Machine 401', 'Apex Copier 347', 'Pioneer Copier 366']
Overlap: 4 of 5

Hierarchies, groups and bins

• Hierarchy: drag Sub-Category onto Category in the Data pane → name it "Product hierarchy". A + on the pill drills down. Date fields have a built-in Year → Quarter → Month → Day hierarchy.
• Group: select marks or right-click a field → Create → Group, e.g. group Indian states into "North-East". A group is a new dimension; its members are fixed.
• Bins: equal-width buckets of a measure (Lesson 2).

Sets

A set is a subset of a dimension that is either IN or OUT.

Use a set on Color to show IN vs OUT, on Filters to keep only members, or in calculations: IF [Top Customers] THEN [Sales] END. Set actions (Lesson 6) let users change a set by clicking marks.

cust = df.groupby("customer_id").agg(sales=("sales", "sum"), profit=("profit", "sum"))
top10 = cust.sales.nlargest(10).index
in_set = cust.index.isin(top10)
print(f"Top 10 customers (IN): {cust.sales[in_set].sum() / cust.sales.sum():.1%} of sales")
unprofitable = cust.index[cust.profit < 0]
print("Customers in Top 10 AND unprofitable (combined set, intersection):", len(set(top10) & set(unprofitable)))
print("All unprofitable customers:", len(unprofitable), "of", len(cust))

Top 10 customers (IN): 5.3% of sales
Customers in Top 10 AND unprofitable (combined set, intersection): 1
All unprofitable customers: 142 of 785

Parameters

A parameter is a single user-controlled value (number, string, date or list) that does nothing until a calculation, filter or reference line uses it. Right-click → Show Parameter.

1. Dynamic Top N

1. Create parameter Top N (integer, range 5–20, step 5).
2. Product Name filter → Top tab → By field → Top [Top N] by SUM(Sales).
3. Show the parameter: users choose 5, 10, 15 or 20.

2. Measure swap

// Parameter "Choose measure": list of strings Sales, Profit, Quantity
CASE [Choose measure]
WHEN "Sales" THEN SUM([Sales])
WHEN "Profit" THEN SUM([Profit])
WHEN "Quantity" THEN SUM([Quantity])
END

Put this calculated field on the axis; one chart now shows whichever measure the user picks. Use a dynamic title: <Parameters.Choose measure> by Region.

3. What-if: extra discount

// Parameter "Extra discount" : float 0.00–0.20
[Sales] * (1 - [Extra discount])                     // projected sales
[Profit] - [Sales] * [Extra discount]                // projected profit

for extra in (0, .05, .10):
    proj_profit = (df.profit - df.sales * extra).sum()
    print(f"extra discount {extra:.0%}: projected profit ₹{proj_profit / 1e7:.2f} crore")

extra discount 0%: projected profit ₹1.84 crore
extra discount 5%: projected profit ₹0.83 crore
extra discount 10%: projected profit ₹-0.18 crore

With margins around 9%, a further 10% discount turns the whole business loss-making. That is exactly the kind of insight a what-if parameter makes obvious to a manager.

Summary