Analytics, Trendlines & Forecast
The Analytics pane in practice.
Trend lines are used to:
- Filter outliers
- Predict continuation of a trend and show correlation
- Change mark colour
- Group dimension members
Answer: Predict continuation of a trend and show correlation — They help identify correlation between variables by showing the trend in both.
Tableau's forecasting model is:
- Linear regression
- Exponential smoothing
- ARIMA
- Random forest
Answer: Exponential smoothing — Tableau uses exponential smoothing to predict future values of a measure.
Forecasting requires:
- A dimension on Colour
- A date dimension and a measure
- An extract
- A dual axis
Answer: A date dimension and a measure — Forecasting works on a time series, so a date field is essential.
A reference line differs from a trend line because:
- It is always horizontal
- It marks a constant or computed value rather than fitting the data
- It cannot be formatted
- It only works on maps
Answer: It marks a constant or computed value rather than fitting the data — Reference lines mark a value such as an average or target; trend lines fit a model to the marks.
Which analytics object shows the spread of the middle 50% of values?
- Trend line
- Box plot
- Forecast
- Reference band
Answer: Box plot — A box plot displays quartiles, with the box covering the interquartile range.
R-squared in a trend line describes:
- The slope
- How much variance the model explains
- The p-value
- The number of marks
Answer: How much variance the model explains — R-squared measures fit; the p-value indicates whether it is statistically significant.
Which chart is best for examining correlation between two measures?
- Pie chart
- Scatter plot
- Gantt chart
- Filled map
Answer: Scatter plot — Scatter plots take two to four measures and reveal relationships.
A histogram requires:
- Two measures
- One measure and a binned dimension
- A geocoded field
- A date
Answer: One measure and a binned dimension — Bins group a measure into bands; the histogram then counts records per band.
Clusters in the Analytics pane group marks by:
- Alphabetical order
- Similarity across chosen measures
- Colour
- Sheet order
Answer: Similarity across chosen measures — Clustering uses k-means over the variables you supply.
Adding a totals row to a view is done from:
- The Marks card
- Analysis → Totals
- The Filters shelf
- Show Me
Answer: Analysis → Totals — Analysis → Totals adds row and column grand totals and subtotals.