I would like to use Orange to create a model that will allow me to predict future WiFi speeds using average quarterly WiFi speed from the last few years. The data sets I am using include the average upload speed, download speed, and lat ms of every country. My project requires that I use three different models. I have chosen to use Linear Regression, kNN, and Neural Network. How would I go about creating this model?
Creating those models should be quite simple. However, you may not have good results with linear regression because this kind of data use to be too complex to be translated in a linear way. You may prefer a logistic regression, with a shorter time frame, starting with one feature.
I recommend using ARIMA, as it could be a good predictor for this case. Therefore, you will want to install the Time-Series add-on (see options, add-ons).
Keep in mind that Orange is very useful to have good data predictions, but there are better predictive models in Python (Multivariate LSTM, Prophet, ...).
Future data set prediction on Orange not possible with supervised ML (and i have not found any useful information how to predict/forecast with unsupervised ML in Orange). Another big lack is Search-Grid feature for no to trial NN's shape and hyper parameters. Basically u re limited to target variable column that u cant surpass. Orange is a nice tool but extremely limited to already known data analysis" only and as such needs a big "upgrade". Use JASP instead, wit feature "Add data prediction" feature (that u don't have in Orange) and u also have a Prophet integrated in 18 version. But JASP features R, Orange features Phyton.