Forecasting Exam #1

Autocorrelation

A measure of the degree of similarity between a given time series and a lagged version of itself over successive time intervals

Standard Deviation

A measure that is used to quantify the amount of variation or dispersion of a set of data values

State whether the data is cross sectional, time series, or panel. Unemployment rate in Minnesota, Wisconsin and Iowa collected ONLY in August 2021.

Cross section: data collected in several units at one point in time.

Why should forecasters know about the decisions that their forecasts might inform. Be specific.

Knowing the decision the forecast might inform helps the forecaster to understand the costs implications of errors in the forecasts. It also helps the forecaster to make decisions about the ideal time series variable to forecasts (for example, percent changes, absolute changes, etc.), understand forecasts horizon, among others.

What is the rationale-philosophy behind Naïve I forecasts?

Naïve I forecast method is the no-value change forecast. It assumes that best guess of what the future value of a time series might be is the actual observed value in the previous (or last period available). This is in contrast to the rationale behind the Naïve II method which takes the changes in the time series observed in the previous two periods and assumes that some of that change will continue to occur into the next period.

State whether the data is cross sectional, time series, or panel. Data on Wages of 5,000 individuals from the US. working population collected for the same individuals in March 2020 and in September of 2020.

Panel data: data collected on the same units across time.

Autocovariance

The covariance between a time series variable and its lagged value. The jth autocovariance of Y is the covariance between Yt and Yt-j.

Variance

The measure of variability, or the degree of spread in the data set

What is the rationale-philosophy behind Seasonal Naïve forecasts?

The seasonal naïve forecast method is an extension of the naïve I method to seasonal data. It is used for purely seasonal time series variables (i.e., only pattern present in the time series is seasonality). This method assumes that the future value at a given time period will be the same as the value observed on the exact seasonal period the previous year. For example, if we are forecasting monthly retail sales for December of 2022, the forecast will correspond to the value of sales observed in December of 2021. Therefore, the implicit assumption is that the seasonal variation will be exactly as the one observed the previous year.

Advantages of objective - Statistical Forecasting

There forecasts are based on large amounts of data and statistical models that can be obtained using computers relatively fast. These forecasts are not subject to judgmental biases and are less likely to be confused with a plan.

Two Disadvantages of Subjective Forecasting

These forecasts can easily be biased because they are based on individual's value judgements. The forecasts are more likely to be subject to peer pressure and strong personalities.

State whether the data is cross sectional, time series, or panel. Payroll employment and the unemployment rate in Minnesota, collected from January 2020 through August 2021

Time series data: data collected on the same unit across time.

What are the statistical characteristics of a time series that has a trend? Why are these characteristics relevant when forecasting a time series? Be specific.

When a time series has a trend it does not have central tendency, meaning the mean is changing across time. Mathematically, we can compute the mean. However, the mean of a time series with a trend should not be interpreted, nor used as a forecast model. Another statistical characteristic of a time series with a trend is that is non-stationary.

What are the differences between univariate and multivariate forecasts? Explain using an example.

When the forecast method only uses the historical values of the time series that is being forecast, the method is called a univariate forecast method. An example is when we forecast a commodity's price (say oil) using the historical observed oil prices. On the other hand, forecast methods that use the historical data of the forecast variable plus additional time series variables are multivariate forecast methods. An example would be forecasting the US unemployment rate using historical data on unemployment, GDP and interest rates.

correlation coefficient

a statistical measure of the extent to which two factors vary together, and thus of how well either factor predicts the other

cross sectional data

data collected at the same or approximately the same point in time

panel data

may be one that follows a given sample of individuals over time and records observations or information on each individual in the sample.

Mean

tells us the central tendency

Covariance

the extent to which two variables are observed to go together Measure of the relationship between two random variables and the extent they change together

time series data

values that correspond to specific measurements taken over a range of time periods