Advanced Analytics AI/ML Features For Designers

Owned by Henry Yiu
Last updated: Aug 30, 2023 by Aadesh Srivastav (Unlicensed)
6 min read

There are several features within Shoplogix Analytics Suite that utilizes machine learning to provide insights. The advanced features described below utilize machine learning. The features are described below with data set requirements in order for the feature to function.

Insight Miner

Feature: The Insight Miner Feature automatically searches for data insights, statistics and correlations within the widget data set.

Insight Miner can be access by clicking on the widget menu (ellipsis) → Get insights

Under Get Insights, the available data point will be listed to run Insight Miner.

Insight Miner Data Set Requirements:

  • Data is less than 10 MB.

  • Must have at least three columns.

  • One column is numeric or binary.

  • Column titles are clear and descriptive.

  • In a utf-8 Unicode format.

  • Indicator widget is not supported.

Explanations

Feature: Explanations will provide possible reasons or correlation for an increase or decrease of a data point in your data set

There are two ways to generate Explanations

  • Directly from a point in a Time Series:

  1. In a time-series widget, right-click the point you would like Explanations for.

  2. Select Explanations.

  • From the Widget Menu

  1. In a widget, click Analyze It → Explanations.

  2. If there are multiple metrics in the widget, select the one for which you would like to generate an Explanation.

Explanations analyzes dimensions and combinations of dimensions of the data set in a time series to compare to a previous point in time. Explanations will tell you which dimension drove the behavior which can be a combination of several dimensions.

Explanations Data Set Requirements:

  • Explanations only works on column, line, area, and bar charts.

  • Explanations only works on time series with a single date dimension.

  • Only certain measures are supported.

    • Aggregate functions (SUM, COUNT) are supported

    • Measures based on other Formulas (AVERAGE, MEDIAN, COUNT) aren’t supported

    • Some custom formulas are supported provided that the formula used is using aggregate functions that will provide the same results in partitioned and non-partitioned data

  • Explanations doesn’t work properly when the measure is filtered by a dependent filter.

  • Explanations can only run on a widget when the relevant table is limited to 115K rows (cannot run if the dataset is larger than 115K rows). This limitation may be solved by applying a widget or dashboard filter to reduce the amount of data in the widget.

Forecasting

Feature: Forecast and predict data points in the future based on the current/past trend

Adding Forecasts to Widgets

  1. Create a widget with one of the supported widget types (Line chart, Area chart, Column chart)

  2. In the Data panel, place a time variable on the X-axis and the time-dependent variable that you want to forecast on the Y-axis

  3. Click the “Owl” icon under the forecasted variable and select Forecast.

 

4. Toggle on the Visible control at the top of the Forecast Settings window to enable the forecast.

5. The read-only forecasted variable is the time-dependent variable that you selected. If you want to add an additional variable, add one in the Explaining Variable field this turn forecast into a multivariate forecast

6. In the Evaluation Period field, select whether you want to use all your past data for the forecast or only a recent period, and select the period.

7. If there is an evaluation period that should be ignored (for example - an unscheduled time period), this can be modified under the Evaluation Period drop down box.

The algorithm requires continuity in data - meaning we cannot exclude a time period in the middle of a time series.

The forecast period defines the number of values to predict.

The AI engine driving this type of forecast uses the following models:

  • Auto Arima

  • Prophet

  • Holt-Winters

  • Random Forest (Random Forest is only used in the ensemble approach; you cannot choose it manually)

These models can be modified in the forecast advanced settings.

To evaluate forecast accuracy:

  1. Once the forecasted period is displayed in the widget, click Analyze It.

  2. Select Forecast → Actual vs. Forecast. The comparison graph appears.

Forecasting Data Set Requirements:

  • Forecasting can be applied to the Line chart, Area chart, Column chart widget types

  • Forecasting requires at least 30 historical data points

  • Designers can add forecasting to widgets, Viewers can then use and customize definitions

  • Must be time based variable (with calendar icon) on the X-axis

    • NOTE - selections such as Date Picker or Shift Instance are not using a calendar type variable and would not be a valid forecasting X-axis selection

  • Y-axis must be a numerical value

  • Data cannot have Break-down selected

Exploration Paths

Feature: Exploration Paths will learn from dashboard usage such as filtering choices and drill down data point selections that are made by the dashboard viewers and designers and will learn from the usage to suggest on what to explore.

Once enough data is collected from the dashboard usage and if there are relevant suggestions to be made; the Exploration Paths feature will display a lightbulb icon on the top right corner of the widget

Enabling AI-driven Exploration Widgets:

  • Exploration Paths has to be enabled per dashboard. Dashboard owners can enable this setting in the dashboard

  • From the Navigation Pane, click on a dashboard to open it or create a new dashboard.

  • Click the ellipsis in the top menu, and select Exploration Paths

 

To know more how these Advanced Analytics AI features algorithms works Click Here!!!

Quest

Feature: Quest allows you to implement advanced analytic models to your widgets, and then add actions that your Viewers can take from the dashboard.

As a Dashboard Designer, you determine what models Viewers can use and what action they perform.

Enabling the Quest for a widget

  1. In your dashboard, edit a supported widget → Go to Design Panel → Click on Quest-Deep Analysis dropdown → Quest Model Editor will be displayed

  2. Enable the models as per your need

3. Click the Model title to display its supported actions, enable any predefined actions.

4. Click “Apply” on widget editor.

For additional information regarding each model,constraints and prerequisites : Click Here!!!