Sign up
Sign in

FOR POWER USER

You have the data, but don't know how to start dealing with the analytical problem or need a baseline model? We've created solutions for industry use cases, like demand forecasting and fraud detection and called them "templates" "template" pipeline is a sequence of bricks needed to get from the data (we've included samples for your convenience!) to the solution, like ML model and inventory forecasting dashboard. It consists of renaming, missing values treatment, aggregation, modeling or prediction bricks. You need to perform a very low amount of operations, to adopt these templates for your datasets and formats. Just connect your data, do minor tweaks to adopt it to the existing pipeline, click on "run" - and you have a fully functioning end-to-end model for your problem, the model could be deployed or dashboard published in a few clicks! Moreover, you can create your template pipelines to share with other power users or viewers.

Enjoy zero- or low-code functionality to do data pre-processing within the datrics.ai platform! We have a lot of bricks to deal with the data through a convenient drag-n-drop user interface. Automated or user-selecter missing values treatment, one-hot encoding, new features creation, filtering, joining and many more operations are on your disposal. Just drag the operation that you need on the left menu and drop it to the main screen, connect with a previous brick output via graphical interface, click on "run" - and enjoy the processed dataset, which could be previewed on each step. Our system will give you a warning or intelligent suggestion if anything goes wrong with any of the operations!

Experimenting: ML or statistical modeling, exploratory analysis

You can upload your own or create the model from scratch in the datrics.ai platform. Do exploratory analysis, see distributions and correlations, build hypotheses and see model performance - you can apply the variety of models via drag-n-drop user interface. From linear regression to boosting algorithms (LGBM), from classification to time series with prophet - select, apply, measure and compare algorithms to select the best candidate for deployment.

Pipeline creation and model evaluation, selection

Drag-n-drop the bricks from data processing, ML/modeling, exporting and visualization/dashboard groups to build a sequence of actions with data without coding. All the phases of a machine learning project are available for you - data upload, cleansing/pre-prosessing, experimenting&modeling, deployment and post-processing, application as a resulting APIs or dashboards creation and export. You can create machine learning models, compare them via plots and relevant statistics, reproduce processing or modeling by creating a pipeline for these purposes.

Collaboration with other power users

The typical problem of Machine learning toolset is a lack of collaborative options and work. Here at datrics.ai you receive it out-of-the-box. Create models or processing pipelines, select part of the pipeline - just save and share them. Re-use common templates in organization to deal with the same problem.

Creating dashboards and APIs (automated) for results sharing and validation

Storytelling is an inevitable part of any ML project. Now it's easy with datrics.ai. Models are deployed in clicks, API is created automatically. Use pre-built or create your own dashboards without coding - share the graphs, summary or data reports with other power users or Viewers (managers, C-level, analysts, marketing team) - whoever requires these! It can be done by sharing the link (just like google docs!) or creating a schedule for pipeline re-run and reports delivery.

Models' performance and versioning

Evaluate the model quality and performance automatically for different datasets. Different statistics and visualization are developed for your disposal, like R2 or error matrix, ROC curve. Deploy and compare the different versions of the same model by using deployment manager.

03

04

05

06

07

Adopting template pipelines for organization's data and needs

01

Data preparation: cleansing and transformation, aggregation and joining

02

FOR VIEWER

Viewing dashboards and saving the results
Do you need the data as a background for decision-making? Are you interested in seeing summary statistics or forecasts to do day-to-day work? The dashboard functionality is here for you. It could be created by Power User or yourself, check it anytime of schedule delivery of the reports.
Do you need to have daily sales forecasts or regular updates on the customer segments? You can have it in datrics.ai by scheduling the reports / pipeline runs delivery, sending the links for shared usage or downloading the report locally. Re-train the model or re-run the pipeline in 1 click!
Change filters, parameters or other relevant configuration to apply your expert knowledge. The set of changeable parameters to be formed by Power User when creating a pipeline, so that Viewer can manually adjust the algorithm inputs.
Interested why the data and results are different from your expectations? Feel free to trace all the way from raw data to the results, which is extremely important for transparency and early problems identification.
Create the different models and use a challenger one in a certain proportion of cases or API calls to measure the real performance. It can be setup from deployment tab (expected in June 2020)

01

Reports and scheduling results delivery and re-runs for new data

02

Viewing the pipelines for transparency purposes

03

A/B testing

04

Configuration changes

05

HOW IT WORKS

Go to Wiki
If you have any question about how to use the product - read our Wiki or email us
Try beta now!

SPECIAL OFFER FOR EARLY ADOPTERS

Be the first among the innovators!
Receive 1 year special price offer for early bird registration!
Try beta - leave feedback - take your discount.
Be the first among the innovators!
Receive 1 year special price offer for early bird registration!
Try beta - leave feedback - take your discount.