Machine Learning

Machine learning is a method of data analysis that automates the construction of analytical models. In order to apply this method with Knolar, it is necessary to follow a series of steps and, before creating the model with which to train our dataset, we must generate the dataset. Once we have our dataset already created, we will proceed to create the model to train the data and be able to detect any anomalies. Finally, we will be able to track the anomalies and visualize the model predictions.

Create Datasets

In order to create a dataset it is necessary to indicate the name and period over which the data will be selected and then add at least one component. These components will be the ones to which we will later apply the model. To add a component to the dataset it is necessary to fill in a series of previous fields (name, ingest, date column, value column, tag column and sensor)

Create Models

Once the dataset has been successfully created, we proceed to the creation of the model. The main purpose of the model is to detect any anomalies in the dataset. To create this model, we must indicate the name of the model, attach a csv example of any event with anomalies and finally indicate the dates on which the Training set, Validation Set and Sampling Rate will be performed.

Monitor

Finally, we will be able to track the different anomalies that our model has detected in the dataset and visualize future predictions.