Forecasting in Tableau - Real Business Scenario

Scenario Mode

👤 Your Role: You are a sales analyst at a retail company.

📋 Request: Your manager wants you to forecast next quarter's sales based on historical monthly sales data.

📊 Data: You have monthly sales data for the past two years, including columns for Month, Year, and Sales Amount.

🎯 Deliverable: Create a Tableau dashboard that shows historical sales trends and a forecast for the next three months.

Sample Data

Year	Month	Sales Amount
2022	January	12000
2022	February	13500
2022	March	15000
2022	April	16000
2022	May	17000
2022	June	16500
2022	July	18000
2022	August	19000
2022	September	20000
2022	October	21000
2022	November	22000
2022	December	23000
2023	January	24000
2023	February	25000
2023	March	26000
2023	April	27000
2023	May	28000
2023	June	29000
2023	July	30000
2023	August	31000
2023	September	32000
2023	October	33000
2023	November	34000
2023	December	35000

1

Step 1: Connect your Tableau workbook to the sales data source containing Year, Month, and Sales Amount columns.

Use 'Text File' or 'Excel' connection to load the data.

Expected Result

Data is loaded and visible in Tableau's Data pane.

2

Step 2: Create a calculated field to combine Year and Month into a date field for proper time series analysis.

DATE(DATEPARSE('yyyy MMMM', STR([Year]) + ' ' + [Month]))

Expected Result

A new date field representing the first day of each month is created.

3

Step 3: Drag the new date field to the Columns shelf and Sales Amount to the Rows shelf to create a line chart showing monthly sales over time.

Columns: Date field; Rows: SUM([Sales Amount])

Expected Result

A line chart showing sales trends from January 2022 to December 2023.

4

Step 4: Enable forecasting by right-clicking on the chart and selecting 'Show Forecast'.

Use Tableau's built-in forecasting feature with default settings.

Expected Result

A forecast line appears extending three months beyond December 2023.

5

Step 5: Customize the forecast to show exactly the next three months and display confidence intervals.

Forecast Options: Set forecast length to 3 months; Show 95% confidence intervals.

Expected Result

Forecast line with shaded confidence bands for January to March 2024.

6

Step 6: Add titles and labels to the dashboard for clarity, including axis titles and a descriptive dashboard title.

Add 'Monthly Sales and Forecast' as dashboard title; Label axes as 'Date' and 'Sales Amount'.

Expected Result

Dashboard is clear and easy to understand.

Final Result

Monthly Sales and Forecast

Date -->
|
|      *
|     * *
|    *   *
|   *     *
|  *       *
| *         *
|*           *
+---------------------->
  Jan22           Mar24

* = Actual Sales
- - - = Forecast with confidence bands

✓Sales have steadily increased month over month for the past two years.

✓The forecast predicts continued growth for the next three months.

✓Confidence intervals show a moderate range of uncertainty but overall positive trend.

Bonus Challenge

Create a parameter to allow the user to select the forecast length dynamically from 1 to 6 months.

Show Hint

Use Tableau's parameter feature to create an integer parameter and update the forecast length setting to use this parameter.

Practice

1. What is the main purpose of forecasting in Tableau?

easy

A. To predict future data points based on historical trends

B. To create static reports without any trend analysis

C. To clean and prepare data for visualization

D. To filter data based on user input

Solution

1. Step 1: Understand forecasting concept

   Forecasting uses past data to estimate future values.

2. Step 2: Identify Tableau's forecasting role

   Tableau applies forecasting models automatically to predict trends.

3. Final Answer:

   To predict future data points based on historical trends -> Option A

4. Quick Check:

   Forecasting = Predict future trends [OK]

Hint: Forecasting always means predicting future values [OK]

Common Mistakes:

• Confusing forecasting with data cleaning
• Thinking forecasting creates static reports
• Assuming forecasting filters data

2. Which of the following is the correct way to add a forecast in Tableau?

easy

A. Apply a filter to the date field

B. Drag the Forecast field from the data pane to the Columns shelf

C. Use the 'Forecast' function in calculated fields

D. Right-click on the view and select 'Add Forecast'

Solution

1. Step 1: Recall Tableau forecast adding method

   Forecasts are added by right-clicking the view and choosing 'Add Forecast'.

2. Step 2: Eliminate incorrect options

   Forecast is not a field to drag or a calculated function; filtering dates doesn't add forecasts.

3. Final Answer:

   Right-click on the view and select 'Add Forecast' -> Option D

4. Quick Check:

   Add Forecast = Right-click menu [OK]

Hint: Add forecast via right-click menu on the chart [OK]

Common Mistakes:

• Trying to drag a non-existent Forecast field
• Using calculated fields for forecasting
• Confusing filters with forecast options

3. Given a time series chart in Tableau with monthly sales data, what will happen if you increase the forecast length from 3 months to 6 months?

medium

A. The forecast will predict sales for 6 months into the future instead of 3

B. The forecast will only show data for the first 3 months

C. The forecast will become less accurate and disappear

D. The forecast will reset to default settings

Solution

1. Step 1: Understand forecast length setting

   Forecast length controls how far into the future Tableau predicts data.

2. Step 2: Effect of increasing forecast length

   Increasing from 3 to 6 months extends the prediction period accordingly.

3. Final Answer:

   The forecast will predict sales for 6 months into the future instead of 3 -> Option A

4. Quick Check:

   Forecast length = prediction period [OK]

Hint: Longer forecast length means longer future prediction [OK]

Common Mistakes:

• Thinking forecast shortens when length increases
• Assuming forecast disappears with longer length
• Believing forecast resets automatically

4. You added a forecast in Tableau but it shows an error message saying 'Insufficient data for forecasting'. What is the most likely cause?

medium

A. The data contains negative values

B. The data has too few time points to create a forecast

C. The date field is not continuous

D. The forecast length is set too short

Solution

1. Step 1: Analyze error message meaning

   'Insufficient data' means not enough historical points to model a forecast.

2. Step 2: Identify common causes

   Too few time points prevent Tableau from calculating trends; other options don't cause this error.

3. Final Answer:

   The data has too few time points to create a forecast -> Option B

4. Quick Check:

   Insufficient data = too few time points [OK]

Hint: Check if time series has enough data points [OK]

Common Mistakes:

• Assuming short forecast length causes error
• Ignoring date field type importance
• Blaming negative values for forecast errors

5. You want to forecast quarterly sales for the next year in Tableau. Your data has monthly sales for 3 years. Which steps should you take to create an accurate forecast?

hard

A. Aggregate data yearly, add forecast for 1 year, and disable forecasting options

B. Use monthly data directly, add forecast for 12 months, and ignore confidence intervals

C. Convert monthly data to quarterly, add forecast for 4 quarters, and check confidence intervals

D. Filter data to last year only, add forecast for 4 quarters, and hide forecast lines

Solution

1. Step 1: Aggregate data to match forecast period

   Since forecasting quarterly sales, convert monthly data to quarterly sums.

2. Step 2: Set forecast length and review intervals

   Add forecast for 4 quarters (1 year) and check confidence intervals for reliability.

3. Final Answer:

   Convert monthly data to quarterly, add forecast for 4 quarters, and check confidence intervals -> Option C

4. Quick Check:

   Match data granularity and forecast length [OK]

Hint: Match data granularity to forecast period [OK]

Common Mistakes:

• Forecasting monthly data for quarterly without aggregation
• Ignoring confidence intervals
• Filtering data too narrowly before forecasting