6.0.2 Released on 8/7/2023
- Much faster previews available for large Excel sheets.
- A variety of additional efficiency improvements and fixes.
6.0.1 Released on 6/5/2023
Data Manager / Insight
- Quick Insights created within the Data Manager can now easily be turned into an Insight saved in the Insight area. When creating a Quick Insight, just click the Save As Insight button to do this.
- Text files generated on MACs (which have different line endings from files created on other systems) will now be automatically identified and correctly imported even if the “MAC File” option wasn’t checked.
- When copying a dataset, if it has an Automated Insights setup, you will be given the option to save those AI settings with the copied version of the dataset.
6.0.0 Released on 4/25/2023
See the LityxIQ 6.0 announcement press release: https://www.prnewswire.com/news-releases/announcing-lityxiq-6-0--powering-predictive-business-decisions-through-next-gen-machine-learning-301806824.html
- ML model settings now allow for easily duplicating and labeling modeling iterations.
- Model Monitoring Dashboards can be created to automatically track in-market model performance, including predictor variable drift and model degradation.
- Enhanced performance analysis to compare many model and versions at once, track overfitting, compare variable importance, and ROC charts.
- New Engagement Analysis model evaluation tool, and enhanced Threshold Analysis with error cost analysis.
- New ability to export detailed data created during the modeling process.
- Ability to automatically put the best model iteration into production for use in scoring jobs.
- Improved coordination of multiple scores being placed into the same scoring catalog.
- Automated Insights settings allow for dataset filtering.
- LityxIQ Shared Insights have been expanded and updated, allowing more flexibility for sharing and managing linked insights that can be made available to non-LityxIQ users.
- Fresh user interface, with main site navigation in a collapsible left panel.
5.3.3 Released on 1/16/2023
- Raw dataset connections to external databases now provide an option to write you own custom SQL query to retrieve data. See https://support.lityxiq.com/848322-Importing-Data-Using-a-Custom-SQL-Query.
- Full projects can now be copied. The copy process can include a full copy of all the data, or be copied without data (just a template of all dataset, insights, and model definitions). Further documentation can be found at https://support.lityxiq.com/237590-Projects.
5.3.2 Released on 12/12/2022
- Improvements to titles and formatting in the multivariate segmentation.
- Changing the name of variables in raw dataset dictionary will not require re-importing the dataset.
- Improvement to date-based scheduling handling when errors occurred in the job.
- Improvement to functionality of user and user-group selection functionality.
5.3.1 Released on 11/7/2022
- The view of multivariate segmentation trees in Automated Insights and decision trees in Predict can now be customized and export to PDF (in addition to PNG and JPG).
- In a derived dataset, Incoming and Join datasets have an option to be inactivated for processing. This is a way to temporarily turn off or pause a dataset from being processed.
5.3.0 Released on 10/17/2022
- Two-factor authentication is now available as an additional security feature on LityxIQ instances. 2FA can be setup by a system administrator to have it enforced for all users on an instance, or can be setup by individual users specifically for their own login. See https://support.lityxiq.com/841701-Enabling-Multi-Factor-Authentication-2FA for more information.
- Two new data connection types are now available - Microsoft OneDrive and a generic REST API connection. The API connectivity allows users to connect to data that is accessible through virtually any API. See https://support.lityxiq.com/484429-Import-Data-from-Microsoft-OneDrive for information on connecting to OneDrive.
- When setting up a new raw dataset, a new option allows users to automatically remove all trailing spaces from strings while importing data from any source.
- Pivot insights provide a more clear understanding of dimensions for which filters have been set, and what the combination of filters is.
- Regression algorithms now provides separate options to search for interaction terms and to identify higher order polynomial terms. The highest polynomial degree for which to search can also be set.
5.2.2 Released on 8/26/2022
- The prior update to New Field Aggregation settings for ranking and N-tiles allowing for the calculation of multiple variables at once will now additionally provide the option to NTile or Rank in descending order as well as specify how null values are sorted. Further, Ntiles can now be computed for up to 1000 groups. See https://support.lityxiq.com/192161-New-Field-Aggregations---Concept-and-Comparisons and https://support.lityxiq.com/081990-New-Field-Aggregations---Function-List
- In Derived datasets with multiple incoming datasets, if the same variable enters with different numeric types (e.g., decimal and integer), the new type will coerce to decimal instead of string.
5.2.1 Released on 8/12/2022
- LityxIQ import file analysis now includes pre-population of raw dataset settings compression, file type, and the delimiter (in the case of delimited files).
- New Field Aggregation settings for functions like cumulative distributions, ranking, and N-tiles will now allow for the calculation of multiple variables at once and the selection of ordering variables will not be forced in situations where they aren’t necessary. See https://support.lityxiq.com/192161-New-Field-Aggregations---Concept-and-Comparisons and https://support.lityxiq.com/081990-New-Field-Aggregations---Function-List for more information on New Field Aggregations.
5.2.0 Released on 8/1/2022
- The K-Means algorithm is now available in beta to support unsupervised clustering or customer segmentation models.
- The Prophet time series algorithm is now available in beta as another option for building forecasting models.
- Scoring catalogs can now be assigned into any dataset library when created.
- Raw datasets in unstructured XML or JSON formats can now be imported. See https://support.lityxiq.com/063632-Import-Data-from-an-XML-File and https://support.lityxiq.com/555837-Import-Data-from-a-JSON-File.
- Datasets can now be Locked. This makes it so the dataset settings can’t be edited and the data itself can’t be modified.
- The data in a dataset can now be "cleared out" using the Clear Data option. This will empty out the dataset without changing anything about its settings or dictionary.
- Administrators will have access to a new System Usage Report which will show how much database volume has been used by the server, and help track the largest datasets.
- Improvements to date-based scheduling processing.
5.1.4 Released on 6/6/2022
Beta Release of Model Explorer
- The Model Explorer beta has additional improvements and enhancements.
- File exports – option to Always Quote all Fields has been added. This can help normalize exports with strange characters in some fields (quotes, carriage returns, etc), but is only necessary if you find you are having trouble importing the export file into another tool.
- Datasets can now be exported to Amazon Redshift.
- When importing data, a new “role” that can be assigned to a field is “PII”, and LityxIQ will automatically detect fields with PII.
- When importing from a zip file, you can now view the files that are embedded within the zip prior to importing.
- Manual datasets are a new kind of dataset that you can create. They are hand-entered datasets, mainly used to enter small datasets manually, or by cutting/pasting data from Excel spreadsheets. See https://support.lityxiq.com/179215-Edit-a-Manual-Dataset for more information.
- Improved decision tree graphic in the Predict area and can be zoomed in/out.
5.1.3 Released on 4/4/2022
Beta Release of Model Explorer
The Model Explorer allows you to dig deeply into the inter-relationships of variables in the model, and their combined effects on model predictions. This opens up greater model interpretability for even the most complex machine learning algorithms like XGBoost and DeepNets. See more documentation on the Analyzer here.
- Users will be informed when a new version has been made available. Just need to refresh the browser using CTRL-SHIFT-R
- Analytic jobs can be canceled in the Queue tab of the console prior to their starting execution, and model runs, scoring jobs, and optimization implementation jobs can also be canceled after they start.
- Right clicking an object in the main grid lists (such as lists of dataset or models) now provides one-click access to that object's menu.
- Small UI fixes and enhancements.
- String fields will be treated differently in dataset dictionaries. The maximum allowed string length is now 65,000 characters, and the length setting in a dataset dictionary will have five options:
- Extra Small (length 1)
- Small (up to 10 characters)
- Medium (up to 100 characters)
- Large (up to 1000 characters)
- Max (up to 65000 characters)
- Additional functionality when creating New Fields in a Derived Dataset:
- You can copy and paste new fields code within the dataset or from one dataset to another.
- The editing window can be sized as needed to show more code, and also has a checkbox to turn off code wrapping for cases where it is helpful to see everything stretched out on a single line.
- When filtering you can now "Check All" to select only those items selected by the filter.
- The drop variables list in the Finalize area now lets you sort the list alphabetically.
- The Multivariable Segmentation in Automated Insights can now be exported.
5.1.2 Released on 8/2/2021
- Login Screen and Welcome Splash updated.
- The table listings in Data Manager (datasets), Insight (charts etc.), Predict (models) and Optimize (scenarios) now use the full real-estate of the screen and any change to the bottom right corner 'Show rows' option (25,50,100 etc.) will be remembered.
- Browse Data can now simultaneously sort multiple columns.
- Browse Data Actions now provide the choice of columns to show, the order to show them and will retain the users settings over time.
- Additional Pivot settings are now saved including 'Analysis Style', 'Aggregation Type', 'Metrics' along with 'Column', 'Row' and 'Unused Dimensions' for the Analysis Style options of Table, Table Bar Chart, Heatmap, Row Heatmap and Col Heatmap.
- Improved error capturing when importing data.
5.1.1 Released on 5/7/2021
- The aggregation step now allows you to pivot data from rows into columns, including summarizing on the pivoted data. This is useful when you want to aggregate your data at a certain level and have a field in your data with multiple values that you would like to turn each value into a new field.
- See LityxIQ Support Site ‘Defining a Pivot within the Aggregation Step’ for more detailed information on how to utilize this new functionality.
- To see this new functionality in action, please view our video demonstration
5.1.0 Released on 1/24/2021
- New aggregation functionality includes creating percent of records and percentage of sums, as well as unique value counts, by aggregation group.
- Options are now available to import SAS, SPSS, or Stata datasets.
- DeepNet algorithm is now fully released
- Jobs will process batches of records in parallel if allowed by the LityxIQ instance setup.
- New option to select the version/iteration of the model to use for the scoring job. Models in production will still always use the official production version. Note that this means that a model not in production can have multiple scoring jobs each using a different version/iteration.
- New option (on the Advanced tab) to allow a scoring job with no records passed in to process without error. Currently, if no records are in the dataset to be scored, the scoring job throws an error. That will continue to be the default operation, but the user can now specify that it is ok for there to be zero records.
- LityxIQ now includes a Gurobi-enhanced upgrade, allowing for solving of Optimization problems using a Gurobi solver.
- Updates and improvements to the user-interface, including detailed user-oriented explanation of constraint definitions.
5.0.4 Released on 9/27/2020
- Additional catching and attempted fixing of import errors, and provision of detailed information on the data line causing any import errors.
- Variable names can now contain up to 128 characters (from 100).
- Imported datasets provided with more than 128 characters in a variable name will be automatically reduced to the first 128 characters to become legal.
- DeepNet neural net algorithm is now available to all users in Beta testing. Please provide feedback on your use of the algorithm in your machine learning models.
5.0.3 Released on 8/31/2020
- Additional algorithm documentation
- The DeepNet neural net algorithm is available to select users in Alpha testing.
- Additional performance metrics are computed for both continuous variable models and binary classification models. See Performance Analysis Metrics to Analyze Classification Models and https://support.lityxiq.com/050028-Performance-Analysis-Metrics-to-Analyze-Numeric-Prediction-Models for a complete list of metrics computed by LityxIQ.
5.0.2 Released on 8/6/2020
- Addition of powerful fuzzy matching functions (in addition to existing regular expression pattern matching functions) to Data Manager.
- An interactive ROC curve and error cost analysis is available for all binary classification models. See https://support.lityxiq.com/364317-C-Statistic-and-ROC-Curves for more information.
- The XGBoost machine learning algorithm is now available to all users
Please do a SHIFT-CTRL-R refresh in your browser to take advantage of the ROC curve analysis functionality.
5.0.1 Released on 5/27/2020
- Automated Insights are now available. For any dataset in LityxIQ, simply select up to five variables for which to generate automated insights each time the data is refreshed. Automated Insights include a ranking of which other variables are important predictors of the target, and provide an easy-to-understand multivariate segmentation for each of the selected targets.
- The export dataset option allows the selection of an escape character for use in the export file.
5.0 Released on 4/1/2020
- Dataset metadata is now computed in parallel to dataset execution being finalized. This provides much-improved execution times for certain large datasets.
- Database schedules now allow for hourly jobs
- Google BigQuery and Amazon DynamoDB are now available as data sources. See Data Connections
- Variable importance scores and other pre-processing information about machine learning model runs is now available for all algorithms