Time-series Forecast Model – Model Results

After you have run your model, let’s see your results. With Model Results, we summarize all relevant results through visualizations and numerical values. First, you can see the values of four evaluation metrics; evaluating a machine learning algorithm is an important part of any project.

R-squared determines the proportion of variance in the dependent variable that can be explained by the independent variable. MAPE (Mean absolute percentage error), MAE (Mean absolute error) and RMSE (Root mean squared error) are measures that describe the average difference between the actual and predicted value.

The results also consist of 5 tabs: Prediction, Trend, Seasonality, Special Dates, and Details. The Prediction Tab contains a graph with actual and predicted values. Besides the main prediction of the target value, the model predicts the range of values for every day in the future, the range between lower and upper value, also known as uncertainty interval. With visualization, you can see how well or poorly your model is performing.

Trend and seasonality are characteristic of time-series data that can be visually identified in time-series plots, so it's important to analyze them too. In the Trend Tab, the graph shows the global trend Graphite detected from the historical data.

Seasonality represents the repeating patterns or cycles of behavior over time. Depending on your Time Interval, in the Seasonality Tab, you can find one/two graphs. In case you have daily data, the first one shows a detected pattern in historical data that repeats every week, while the second refers to a year. For example, in the picture above, you can see that the biggest positive impact happens on Mondays. For weekly and monthly data, the graph shows detected patterns in data that repeat every week/month through the year.

In the end, a table with all the values ​​related to the Prediction Tab, with much more, can be found on the Details Tab. Later, we will talk about the Special Dates Tab. By running this model, you get fast and quality insights into your business - just one new piece of information obtained from the data can considerably help in further business. Now it's your turn to do some modeling and explore your results. Enjoy! 🙂