Explore our features

Features

Machine Learning

Features

Machine Learning

The Machine Learning module in Dashtera enables users to build predictive models using supervised learning techniques for both Regression and Classification tasks. Dashtera includes eight machine learning algorithms designed for real-world prediction problems. Users can train models, evaluate results, store trained models, and run predictions on new datasets without writing any code.

Dashtera currently supports two ML task types:

Regression – used to predict continuous numerical values

Classification – used to predict categorical labels/classes

Supported Models

Dashtera provides eight supervised learning models, grouped into two categories:

Regression Models

Linear Regression
Random Forest (Regression)
LightGBM (Regression)
XGBoost (Regression)

Classification Models

Logistic Regression
Random Forest (Classification)
LightGBM (Classification)
XGBoost (Classification)

These models accommodate a wide variety of use cases, including forecasting, pattern recognition, and decision-based predictions

Machine Learning Workflow

The Machine Learning module is organized into three main tabs, each designed to guide users smoothly through the full modeling process – from training to prediction.

Training Module

The Training Module allows users to build machine learning models through a guided and user-friendly process. A dataset can be selected directly from Dashtera Data Sources, after which the target variable and desired feature columns can be chosen for training.

Users have the flexibility to select Auto Train, where Dashtera automatically selects the most suitable model for the dataset, or manually pick a model from the dropdown list. The model list dynamically changes based on the selected task type – Regression or Classification.

Once the configuration is complete, pressing Analyze generates a performance summary including metrics such as Accuracy, Precision, Recall, F1-Score, Train/Test Scores and more, helping users validate the effectiveness of the model before saving.

Training Steps:

Select a dataset from Data Sources
Choose the target variable
Select/deselect feature columns
Enable Auto Train (Dashtera picks the best model) – or manually select a model from the dropdown (Model list updates automatically based on task type)
Click Analyze to generate training results and performance metrics
Save your trained model with a custom name

This structured flow ensures fast model development and easy experimentation, even for non-technical users.

Model Repository

The Model Repository stores trained models along with their details such as Model Type, Target Column, Feature Columns, Scores, and Saved Time. Up to five models can be stored, allowing users to maintain multiple versions or experiment with different configurations. Saved models can be reused directly for future predictions.

Prediction Engine

The Prediction Engine enables users to run predictions using any model stored in the repository. By selecting a dataset and choosing the desired model, predictions can be generated instantly. Resulting data is saved back into Dashtera Data Sources and can also be exported for external use.
Supported export formats:

Excel
CSV
JSON

This tab completes the ML workflow, taking models from training to deployment-ready prediction output.

Machine Learning in Dashtera simplifies AI adoption for organizations by bringing model training, evaluation, management, and deployment into one unified no-code platform. Users can seamlessly move from raw data to predictive insights without external tools or programming expertise.

Once predictions are generated, the results are stored back into Dashtera Data Sources, allowing users to instantly visualize outcomes, create reports, and build interactive dashboards through Dashtera BI. This enables transitioning from predictive analytics to decision-making, performance tracking, and business intelligence dashboards, all within the same ecosystem. Dashtera connects the full workflow end-to-end

Machine Learning Sales Prediction Model Example

Below is an example of how Dashtera Machine Learning can be applied to real business scenarios. A Sales Forecast Model was trained using Linear Regression to predict monthly sales based on price, discount levels, advertising expenditure, store size, and region.

The model achieved an impressive 97.04% Train/Test Score, meaning it explains most of the variation in sales data accurately. The chart below shows the comparison between Actual Sales (solid line) and Predicted Sales (dashed line).

Both lines follow a similar trend over time, showing that the model closely matches real performance. When they move slightly apart, it reflects months with unusual spikes or dips. Overall, this visualization demonstrates how effectively Dashtera forecasts sales trends and supports future planning.

Sample records Used for Training

Month	Price	Discount	Ad Spend	Store Size	Region	Actual Sales	Predicted Sales
2021-01	131	4	5737	Small	West	605	544.51
2021-02	94	25	1854	Large	West	536	527.48
2021-03	140	22	9164	Small	South	748	724.19
2021-04	100	13	6855	Small	East	712	678.02
2021-05	103	6	5392	Small	East	738	729.28