BUS 346 Lecture Notes - Lecture 3: Moving Average, Exponential Smoothing, Average Absolute Deviation
12 Sep 2017
School
Department
Course
Professor
Document Summary
Match supply with demand so that we don"t have excess or shortage. Why don"t we just guess: accuracy matters helps balance supply and demand better, hard to sort through the noise. We often get distracted by points that are far from the trend which could lead to major downfalls. Forecast error = diff b/w actual val and predicted value. Error = actual forecast: et = at ft e = error, a = actual val, f = forecast; t = given time period. Mad (mean absolute deviation) errors are weighted evenly. Mse (mean squared error) errors weighted according to their squared values: used more than mad, problem is that it"s dependent on units. Get a number that you won"t be able to interpret very well. Mape (mean absolute percentage error) errors weighted by relative errors. Mean error would use it to determine whether there is a bias: if it"s really high/low number then it means that there"s probably bias.
