How does this work technically?
Step 1: Training the algorithm
Through “Feature Selection” and “Feature extraction” algorithms, which are executed completely autonomously, Trendskout independently detects which set of variables – eg. columns, pixels … – must be used in the classification model.
During training, Trendskout evaluates various algorithms, algorithm parameters and data transformations. To evaluate this, different quality scores are calculated by running the model on an unused part of the training data and testing whether the discovered relationships are correct. Based on the result of this evaluation, Trendskout will independently try new combinations of algorithms, parameters and data transformations. This part of the training step is called hypertuning. This entire process is started in Trendskout via a simple click on the Run / Train button and is done without user intervention.
Step 2: Real-time classification
The best classification model is then used in the second step; classifying new data in real time. Clicking on Deploy will open the classification model for a data input method chosen in the Connect step. This can be based on API, Plugin, database or others. Each time, Trendskout will classify new data in these input sources, based on the classification model, and perform the selected actions in the Automate step, e.g. calling an external system, sending an email, showing a dialogue …
In this second phase, Trendskout will also adjust and evaluate the previously prepared model. As soon as a different algorithm / data transformation / parameter combination yields a better model, the previous model is replaced. This guarantees the best result, even when the properties of your data change.