Reviewing Analytics Workspace

Analytics Workspace has the ability to use analytics capabilities like scatterplot, data highlighting, significance testing, predictive models, and career trajectory to identify critical workforce issues. Ability to design initiatives based on evidence helps to improve the return on investment of HR activities.

Analytics Workspace consists of 5 tools:

• Scatterplot - build a scatterplot to compare two time periods or metrics, and visualize the magnitude.
• Data Highlighting - build a data highlighting investigation to automatically find problem areas or outliers.
• Significance - test the statistical significance of a metric result.
• Predictive - build a predictive model to be applied to a workforce plan.
• Trajectory - visualize the career trajectory of your workforce.

Compare two time periods or metrics and visualize the magnitude.

• Users have the ability to quantify a relationship between two measures, or one measure over two time periods, whilst also analyzing the strength of that relationship through correlation and regression.
• Users can exclude outliers based on standard deviations, show benchmarks and targets.
• Users can zoom to focus on a desired area.
• Users can include in a report.

Scatter plots are used in conducting correlation comparisons between either two measures over a single time period or the one measure compared in two different time periods. Depending on the data spread, there can be a positive or negative correlation relationship between the two variables. If the data slopes downwards there is a negative relationship and a positive one if the data slopes upward. For instance, in the scatter plot shown you would predict a positive correlation between total separation rate and voluntary separation rate, as the data appears to have an upward slope. Correlation relationships such as these can be useful in determining factors which may affect a particular measure or KPI.

You could look at the relationship between Staffing Rate - High Performers and Customer Satisfaction to see if there is any statistical correlation between these two measures. Each dot represents a business unit, and the size of the dot is determined by the measure EOP Headcount. The results are further broken down into three series based on geographic location, which allows for deeper analysis to be conducted.

There are several areas to configure for a scatterplot:

With the options menu, you can:

• Create a new scatterplot session.
• Select a previously saved session to edit.
• Save the current session.
• Copy the current session to edit.
• Share the scatterplot session with Page Designer.
• Open a help box detailing the options in the file menu.

Query type: Choose between comparing one measure over two time periods, or two measures in one time period.

For two measures and one time dimension:

1. Select the Measures for the x and y axes.
2. Select the Time period.
3. Select the Dimensions and Level within the structure.
4. Apply filters (optional).
5. Select Benchmarks (optional).

For two measures and one time dimension:

1. Select the Measure to compare.
2. Select the Time periods on the x and y axes.
3. Select the Dimensions and Level within the structure.
4. Apply filters (optional).
5. Select Benchmarks (optional).

• Toggle outlier removal on/off.
• Select the number of standard deviations away from the mean outliers need to be removed.

• Exclude points with a zero value on the x axis.
• Exclude points with a zero value on the y axis.

• Toggle the regression line on/off.
• Toggle the line equation on/off.
• Toggle the correlation coefficient on/off.
• Toggle the regression line per series on/off (if included a ‘Breakdown Dimension’).
• Indicator type of None, Target, or Benchmark. Indicator is a background to the scatterplot.

The zoom function enables users to focus on a particular part of the graph. To return to the full view, select the Reset button.

Find problem areas or outliers.

• Join together a lot of different criteria to scan across large datasets.
• Users can include in a report.

Data highlighting combines one or more expressions to construct a table in order to identify very specific examples. Large data sets can be analyzed for specific cases.

You could look for divisions with over 100 staff and where you primarily recruit externally how successful we are at retaining individuals in critical job roles in their first few years of employment. This could also focus on short tenure voluntary terminations where the divisional average is greater than 25%.

This could provide a list of divisions that have many external hires for critical roles but did not retain the talent. This could be an important factor because of the high cost of hiring externally.

Data Highlighting Configuration

To start building a search, select the Add a new expression item in the Options menu. This will create the initial criterion that defines the data you are wanting to locate. A more complex search can be created by adding subsequent criteria. Each search criterion is referred to as a 'search expression'. The completed set of search criteria is called a 'query'. Once the search expressions are defined, select Run Query for the current query to show the results.

Add an Expression

Use the Options menu.

Select Measures and Define Search Parameters

When defining the search parameter, select where and comparison operator.

For where:

• Values lie in a benchmark range of an internal benchmark: This option will sort all the measure values for the selected dimension nodes and search for those nodes that fit within the percentile values stated in the expression, for example, if 10 dimension nodes are selected in the query and the search is to find nodes greater than 50%, then the top 5 nodes will be returned.
• Values lie in a benchmark range of the program benchmark: This option compares the measure values for the selected dimension nodes against the selected program benchmark, and searches for those nodes that fit within the percentile values stated in the expression.

Select Reporting Unit

Search Options:

• Search the whole structure: This selects all nodes, both parent and children.
• Search end-points only. Don’t include parent elements: This only selects the bottom end of the structure.
• I want to select a customized set of elements: This allows you to examine specific elements and nodes of your choosing.

Using Custom Set of Elements

The process of defining this set starts by choosing an element from the explorer at the bottom of this view. The selected element will become a reference to which other elements may be added. The elements to be added will be based on the selection made from the menu which appears when you select the Action button.

An important concept in interpreting some of the available actions is distance. This refers to the arithmetic difference in depth between the reference element and the associated target element(s). For example, if you wanted to examine the items immediately below reference element (the child elements), then a distance of one (1) is required. Similarly, the grand-children elements are specified by a distance of two (2).

The following provides more detail about the options available from the action menu:

• Add the selected element and all elements below to the specified distance: This option requires a valid number to be entered in the distance field. It will result in the selection of the reference element plus all the elements below, down to a level specified by the distance value.
• Add all elements below the selected element to the specified distance: This is the same as above but does not include the reference element.
• Add all elements below the selected element which have no elements below them: This option requires a valid number to be entered in the distance field. It will result in the selection of the elements below the reference element at a level specified by the distance value. It does not include the reference element, nor does it include any elements at depths between the reference element depth and the target depth.
• Add all elements below the selected element at the specified distance only: Includes all the elements below the selected element which have no 'child' elements. These terminal or leaf elements have no elements existing below them.
• Add the selected element and all elements below the selected element: Includes the reference element and all elements below it.
• Add all elements below the selected element: Same as above except it does not include the reference element.
• Delete the selected item from the list above.

Select Time Period

Note

If you selected to compare two time periods when you were assigning the metric, you will get a ‘time period to search’ and a ‘time period to compare’.

Filter Results

If required, you can restrict the data to be searched by nominating one or more criteria to which the search shall be limited.

Note

Filters are cumulative. Each additional filter will make the search area successively smaller.

Add Additional Expressions

Add additional expressions to further refine you query. You can add more than one expression.

Edit or Delete an Expression

Select Edit Expression to return to that expression or Delete Expression from Query.

Filter Options

Filter to only display a top or bottom number of records.

Manage Saved Data Highlight Queries

The completed query can be saved for later use if desired. To achieve this, there are two options in the Edit menu that allow you to enter a name that you can use to recognize the query in the future. The queries that you may have stored earlier can be accessed by selecting the Open a previously saved query item in the Edit menu. These queries can be edited and re-saved (under the same name, using Save the current query) or modified and stored as a new query (under a different name, using Save a copy of the current query with a new name).

Test the statistical significance of a metric result.

• Determine the likelihood a value is different due to chance.
• Adjust the level of significance percentage.
• Support breakdowns.
• Users can include in a report.

Used to determine if the results of an analysis by dimensions is statistically significant. The chi squared test determines the probability that the difference between dimension nodes is due to chance or is dependent on the selected node of the dimension.

You might be interested in low tenured turnover within your organization. The graph highlights the proportion of total employee initiated terminations involving employees with less than one year’s tenure. It appears that a number of business units have high rates of low tenured turnover. You are interested in determining whether or not less than one year tenure turnover as a proportion of all tenured turnover is significant. In this example, they are significant, but you will also want to understand if that is significant across selected business units.

Build a predictive model to be applied to a workforce plan.

• Support for filtering of metrics of dependent and independent variables.
• Controllable time granularity.
• Preview Mode to compare changes to the current model.

Used to determine if a particular metric is a driver of productivity. The independent variable will be used to derive a predicted value which can be compared with actual results. If the values are close, you can determine a critical measure to predicting productivity.

This chart predicts whether Operating Profit can be explained by the changing population in a critical job role. The historic trend for Operating Profit is indicated by the blue line. The green line indicates the predicted value for Operating Profit based on the change in the headcount in critical roles. The closer the two lines are to each other the closer the relationship between the two variables.

This type of analysis allows you to determine what the key drivers of organizational outputs (in this case revenue) are. Understanding these relationships allows you to make more confident predictions around future demand forecasts.

Visualize the career trajectory of the workforce.

• Visualize groups or individual employees.
• Configure a target zone.
• Perform regression.
• Users can include in a report.

Visualize career trajectory of the workforce by group or individual. Magnitudes can help visualize the difference in desired and actual outcomes.

You may want to look at what the optimal career promotion path is based on organizational tenure for managers. You may also want to understand how this varies by the number of people in each. The example shows managerial job grades from 1 up to 8. The shaded target zone shows the optimal timeframe over which you want individuals to advance to the M4 level. So, in the best case, individuals would reach that level in 5 years and in the least optimal case in 8 years. Overlaying this target area on the actual data points allows us to analyze actual promotion trends in the workforce by grade and location.