ECON

Problem Definition and Data Collection

• define objectives and constraints.

• Identify variables and time horizon.

• Gather historical data.

Exploratory Data Analysis (EDA)

• Visualize data (plots, charts).

• Identify patterns, trends, and seasonality.

• Calculate summary statistics.

Model Selection and Fitting

• Choose appropriate forecasting method.

• Fit model to historical data.

• Estimate model parameters.

Forecast Generation

• Apply chosen method to make predictions.

• Generate point and interval forecasts.

• Consider uncertainty in forecasts.

Evaluation and Monitoring

• Evaluate forecast accuracy (MAE, MSE).

• Monitor performance over time.

• Adjust models as needed.

White Noise

• definition: A time series where observations are independent and identically distributed with a constant mean and variance.

• Characteristics: Random fluctuations with no discernible pattern or correlation.

• Properties: Mean is constant over time; autocorrelation is zero for all lags except at lag 0.

• Significance: Used as a baseline for comparing the performance of forecasting models.

• Example: Random variations in daily stock prices with no clear trend or seasonality.

Seasonal Adjustment

• Definition: Removing predictable seasonal patterns from time series data to isolate underlying trends and irregularities.

• Purpose: Enhances analysis by focusing on non-seasonal variations, improving forecasting accuracy.

• Methods: Include seasonal decomposition and adjustment factors.

• Importance: Reveals true underlying behavior of data.

• Example: Adjusting monthly retail sales data to remove holiday shopping peaks for a clearer trend analysis.

Decomposition

• Definition: Breaking down a time series into its components—trend, seasonal, and irregular—to better understand underlying patterns.

• Purpose: Facilitates analysis by isolating distinct components, aiding in forecasting and anomaly detection.

• Methods: Multiplicative or additive decomposition.

• Importance: Helps identify long-term trends, seasonal fluctuations, and irregular fluctuations.

• Example: Breaking down monthly sales data into trend, seasonal variations, and random fluctuations for more accurate forecasting.

• This is what you use when first looking at a graph, look to see seasonality, explain the overall trend, and then identify where in the data the irregular points occur.

Order of 5 forecasting steps

Problem definition and data collection

Exploratory Data Analysis

Model Selection and Fitting

Forecast Generation

Evaluation and Maintenance