Timeseries Forecast Model is a machine learning model which predicts future values based on previously observed time-related values. The model can only be run on a dataset that contains time-related and numeric columns. To get the best possible results, we will go through the basics of the Model Scenario. In Model Scenario, you select parameters related to the dataset and model.
For the Time/Date Column, select the column name that contains time-related values. Time Interval represents the frequency of the data. For example, if you have daily data, select daily, or if you have annual data, select yearly. After that, you have to select the Target Column, a numeric value that you want to predict. It is essential to have values by day, week, or year; if some dates are repeated, we can take their sum, average, etc. With Forecast Horizon, you choose how many days, weeks, or years you want to predict (from the last date in the dataset).
The model gives excellent results for data with seasonal patterns. If your data shows linear growth trend, select additive for Seasonality Mode, or if data shows exponential growth trend, select multiplicative. For example, if you notice the same behavior on an annual basis, you can set the Yearly Seasonality as True (TIP: it is useful to plot the data before modeling to better understanding ). If you're not sure, don't worry, the model will try to automatically detect seasonality if there are any.
Once you have selected all the fields, select Run Scenario to run the model. Forecasting is valuable to businesses because it gives the ability to make quality strategies and business decisions. Forecasting helps you plan the future and anticipate changes in the market. With our model, you can make accurate predictions that will have a positive impact on your business! 🙂