3 Pronósticos y Patrones de las Series

Understanding Forecasting

Importance of Forecasting

• Forecasting is essential for estimating the future behavior of a variable of interest, helping in decision-making processes.

• In production management, forecasting helps understand expected demand for products based on various inputs.

Inputs for Forecasting

• Inputs can include market research, historical demand data, advertising campaigns, and expert opinions when market behavior is unknown.

• Understanding future product behavior allows better planning of limited resources such as finances and production capacity to meet expected demand.

Types of Forecasting Models

Qualitative vs. Quantitative Methods

• Forecasting methods are categorized into qualitative (based on human judgment) and quantitative (based on historical data analysis).

• Qualitative methods may involve gathering insights from sales teams or experts about future product sales.

Focus on Quantitative Methods

• The course will focus on objective or quantitative forecasting methods that analyze historical data to characterize variables.

• Causal models use data from different sources to forecast macroeconomic variables but will not be covered in this module; instead, time series models will be used.

Time Series Models

Historical Data Utilization

• Time series models rely on historical data to predict future values of a variable by analyzing past trends.

Identifying Patterns in Time Series

• Recognizing patterns in time series is crucial; four main patterns include stationary demand, trend demand, seasonal demand, and cyclical variations.

Stationary Demand Pattern

• Stationary demand fluctuates around a central value without significant changes over time.

Trend Demand Pattern

• Trend demand shows a linear pattern where sales can either increase or decrease over time based on market performance.

Seasonal Demand Pattern

Understanding Demand Patterns in Time Series Forecasting

Seasonal Demand and Trends

• The discussion begins with the observation that electricity demand is low in the morning but increases significantly at night, indicating a cyclical pattern of demand that repeats daily.

• It is noted that some products may exhibit similar cyclical behavior while also showing seasonal increases from one period to another, highlighting the importance of analyzing these patterns for forecasting.

Identifying Time Series Patterns

• The speaker emphasizes the necessity of graphing data when approaching forecasting exercises to identify recognizable patterns within time series data.

• For stationary time series with little variability, simple moving averages and simple exponential smoothing are recommended as effective forecasting methods.

Advanced Forecasting Techniques

• When dealing with time series exhibiting trends (either increasing or decreasing), linear regression and double exponential smoothing are suggested as suitable forecasting techniques.