Forecasting Models - Explanation of Forecasting Models

Here are the four forecasting models that are available and a brief description of how they work. We recommend keeping the default Ensemble model.

Ensemble

Default model (Recommended)

This model will automatically run 4 different models (Auto ARIMA, Prophet, Holt-Winters, and Random Forrest) on the historical data and compares the output with historical data and will select the model or blend the models for the best result.

Auto Arima

How Does It Work?

How Auto ARIMA Works
Study Past Data: It looks at your historical numbers.
Find Patterns: Identifies trends (up/down) and seasonality (regular cycles).
Test Formulas: Tries different math models to see which works best.
Choose the Best: Picks the formula that fits your data.
Predict Future: Uses that formula to forecast future numbers.

Holt-Winters

How Does it Work?
Smooth the Data: It averages out ups and downs to see clearer trends.
Account for Trends: Adjusts for gradual increases or decreases over time.
Handle Seasonality: Adds regular patterns (like monthly or yearly cycles).
Combine Everything: Uses all these pieces to forecast future numbers.

Prophet

How Does it Work?
Look at Past Data: It analyzes your historical numbers.
Break Down Patterns: Separates the data into:
Trend: Long-term increase or decrease.
Seasonality: Regular cycles (daily, weekly, yearly).
Special Events: Handles holidays or unusual spikes.
Handle Missing Data: It’s good at dealing with gaps or irregular data.
Combine Everything: Puts all these pieces together to forecast future numbers.