KNN-WG

Weather prediction requires extensive statistical information and the knowledge and ability to recognize patterns in large data sets. By doing so, specialists generate future data and can make assumptions about the upcoming weather conditions.

One of the most simple techniques used in this field is the K-nearest neighbors (K-NN) method, which assumes that the climatic conditions are replicating. KNN-WG is a software solution that applies this technique, allowing researchers to run the model and perform the simulation in a digital environment.

KNN-WG displays a fairly simple interface that features a few tabs to help you handle the data and apply the K-NN technique. It can extract information from Excel files, both in XLS and XLSX format. Unfortunately, there is no other supported format as input. The data is then displayed within the main window.

The application works with various weather variables, including the maximum and the minimum temperatures, the rain amount, the humidity, Srad, ETo, and the wind speed. You have to select the corresponding column for each of the values manually, although it would probably be easier if you could configure the structure of the loaded spreadsheet while importing it.

The data can be then loaded as the input for the K-NN model, which can be run in the next tab. First, you have to set the base and the future periods of time, considering that this approach relies on the assumption that current weather data is a copy of past data.

KNN-WG displays the output data for all the variables you have chosen, revealing its daily predictions. It also generates an output plot for a more representative data visualization. The application calculates efficiency criteria and can compare the results of K-NN to those of other models.