Table of Contents

Enum MultiplicativeAlgorithmType

Namespace
AiDotNet.Enums.AlgorithmTypes
Assembly
AiDotNet.dll

Represents different multiplicative algorithm types for time series analysis and forecasting.

public enum MultiplicativeAlgorithmType

Fields

GeometricMovingAverage = 0

Uses a Geometric Moving Average to analyze and forecast time series data.

For Beginners: A Geometric Moving Average (GMA) is like a regular moving average, but instead of adding values and dividing (arithmetic mean), it multiplies values and takes the nth root (geometric mean).

Imagine you're tracking the growth of an investment:

  • A regular average might tell you the average value over time
  • A geometric average tells you the consistent growth rate that would achieve the same final result

For example, if an investment grows by 10% one year and 20% the next:

  • The arithmetic average is (10% + 20%)/2 = 15%
  • The geometric average is v(1.10 × 1.20) - 1 = 14.89%

The geometric average is slightly lower but more accurate for compounding growth.

The Geometric Moving Average:

  1. Is better than simple averages for data with growth rates (like financial returns)

  2. Reduces the impact of outliers and volatility

  3. Preserves the multiplicative relationships in the data

  4. Is commonly used in financial analysis and stock market technical indicators

In machine learning applications, GMA is useful for preprocessing financial time series data, analyzing growth patterns, or creating features that capture multiplicative trends.

LogTransformedSTL = 2

Uses a log-transformed Seasonal and Trend decomposition using Loess (STL) to analyze time series data.

For Beginners: Log-Transformed STL is a technique that first converts your data using logarithms, then applies a powerful decomposition method to separate different patterns in your time series.

Imagine you have a photo that contains multiple layers (background, people, objects). STL is like a tool that can separate these layers. The log transformation makes it easier to separate these layers when they have multiplicative relationships.

The process works in two main steps:

  1. Transform the data by taking the logarithm of each value
  2. Apply STL (Seasonal and Trend decomposition using Loess) to break down the transformed data into:
    • Trend component (long-term direction)
    • Seasonal component (repeating patterns)
    • Remainder component (what's left after removing trend and seasonality)

After analysis, you can transform back to the original scale using exponentiation (the opposite of logarithm).

Log-Transformed STL:

  1. Handles complex seasonal patterns that can change over time

  2. Is robust against outliers (unusual data points)

  3. Works well for data with multiplicative relationships between components

  4. Provides a flexible way to decompose time series with multiple seasonal patterns

  5. "Loess" refers to a special smoothing technique used in the decomposition

In machine learning applications, this method is valuable for preprocessing time series data before feeding it into other algorithms, for anomaly detection (finding unusual patterns), or for understanding the underlying components driving your time series.

MultiplicativeExponentialSmoothing = 1

Uses Multiplicative Exponential Smoothing to analyze and forecast time series data.

For Beginners: Multiplicative Exponential Smoothing is a forecasting method that works well for data with trends and seasonal patterns that change proportionally to the overall level.

Imagine a retail store where sales increase during holidays. If overall sales double over five years, the holiday boost might also double (from +$5,000 to +$10,000). This is a multiplicative pattern.

This method breaks down your data into three components:

  1. Level (the base value)
  2. Trend (the overall direction)
  3. Seasonality (repeating patterns)

But instead of adding these components (Level + Trend + Seasonality), it multiplies them (Level × Trend × Seasonality).

Multiplicative Exponential Smoothing:

  1. Works well when seasonal variations increase as the trend increases

  2. Uses "smoothing parameters" that determine how quickly the model adapts to changes

  3. Gives more weight to recent observations and less weight to older ones (that's the "exponential" part)

  4. Is also known as Holt-Winters multiplicative method

In machine learning and forecasting, this method is particularly useful for sales forecasting, demand planning, stock market analysis, and any time series where the seasonal variation is proportional to the level of the series.

Remarks

For Beginners: Multiplicative algorithms are special methods used when analyzing data that changes over time (time series data), especially when the pattern of change depends on the current value.

Think about the difference between:

  1. Adding $100 to your savings each month (additive growth)
  2. Growing your savings by 5% each month (multiplicative growth)

With multiplicative patterns, the changes get larger as the base value gets larger. For example, 5% of $1000 is $50, but 5% of $10,000 is $500 - the same percentage creates bigger absolute changes as the value grows.

Multiplicative algorithms are especially useful for:

  1. Financial data (stock prices, sales figures)
  2. Population growth
  3. Seasonal patterns that grow or shrink proportionally to the overall trend
  4. Any data where percentage changes are more important than absolute changes

In contrast to additive methods (which use addition and subtraction), multiplicative methods use multiplication and division to model changes. They often work with data on a logarithmic scale or with ratios rather than differences.

This enum specifies which specific multiplicative algorithm to use for analyzing or forecasting time series data.

Edit this page